# Portal Scanner Configuration
# Used by /career-ops scan to discover new offers automatically.
#
# Strategy (4 levels):
#   0. Local parser → per-company `parser:` script for SSR/static career pages (zero-token)
#   1. Playwright → careers_url of each company (real-time, reliable)
#   2. HTTP JSON (WebFetch) → per-company `api:` on tracked_companies (see board_api_catalog below)
#   3. WebSearch with site: filters (broad discovery, may be stale)
#
# IMPORTANT: Each company in tracked_companies MUST have careers_url.
#   Prefer the BRANDED company careers URL whenever available; use ATS-hosted URL only as fallback.
#   ✅ https://careers.mastercard.com   ⚠️ https://mastercard.wd1.myworkdayjobs.com (fallback only)
#   ✅ https://openai.com/careers       ⚠️ https://job-boards.greenhouse.io/openai (fallback only)
#   Exception: ATS-hosted careers_url is acceptable when no branded careers page exists.
#   Using raw ATS URLs when a branded page exists can cause false 410 errors (job ID mismatch).
#
# HOW TO CUSTOMIZE:
#   1. Copy this file to portals.yml in the project root
#   2. Edit title_filter.positive with YOUR target role keywords
#   3. Add/remove companies in tracked_companies
#   4. Adjust search_queries for your preferred job boards
#   5. Set enabled: false on companies you don't care about

# -- Scan history recheck policy (optional) --
# By default, URLs recorded in data/scan-history.tsv are deduped forever.
# Set recheck_after_days to let previously added URLs become eligible again
# after N days, useful for roles that reopen or are republished at the same URL.
# Permanent invalid statuses such as blocked host or malformed URL still remain
# deduped forever.
#
# scan_history:
#   recheck_after_days: 30

# -- Location filter (optional) --
# Filter scanned jobs by location. Applied AFTER title filter, BEFORE dedup.
# If this entire block is absent, all locations pass (current default behavior).
#
# Semantics (case-insensitive substring, in this order):
#   - Empty location string on a job → pass (don't penalize missing data)
#   - Any `always_allow` keyword present → pass (takes precedence over `block`)
#   - Any `block` keyword present → reject
#   - `allow` empty → pass (already cleared block)
#   - `allow` non-empty → must match at least one keyword
#
# `always_allow` is optional. It rescues multi-location postings that name your
# home region: with always_allow ["New York"] and block ["India"], a job listed
# "Remote, New York or India" passes (New York wins), while "Remote, India" is
# still rejected. Omit always_allow entirely and the filter behaves as before.
#
# Example below targets US-based remote + a couple of US metros, blocking
# common foreign hubs. Customize to your geography.

# location_filter:
#   # always_allow is checked BEFORE block — keep this list to your home region only
#   always_allow:
#     - "United States"   # (replace with your home region)
#     - "New York"
#   allow:
#     - "Remote"
#     - "United States"
#     - "USA"
#     - "Atlanta"
#     - "New York"
#   block:
#     - "India"
#     - "Bengaluru"
#     - "United Kingdom"
#     - "London"
#     - "Germany"
#     - "France"
#     - "Singapore"
#     - "Japan"
#     - "Brazil"
#     - "Australia"

# -- Salary filter (optional) --
# Filter scanned jobs by annual compensation range. Applied AFTER location
# filter, BEFORE dedup. Only applies when structured salary data exists.
#
# Semantics:
#   - min/max are annual compensation filters (use annualized values)
#   - max: 0 means "no upper limit"
#   - If no salary data exists on a job, it passes (conservative behavior)
#   - If both currencies are known and differ, the job is rejected by the salary filter
#   - Partial ranges (min only or max only) work correctly
#   - If this entire block is absent, salary is not filtered (default behavior)
#
# Example: target roles paying at least $100k, no upper limit
# salary_filter:
#   min: 100000
#   max: 0
#   currency: "USD"

# -- Posting-age filter (optional) --
# Opt-in freshness gate. Skips offers whose posting date (postedAt, supplied by
# the provider) is older than N days. Offers from providers that expose no date
# are never skipped — same "don't penalize missing data" rule as location_filter.
# Absent, 0, or non-positive → disabled (all ages pass).
#
# max_posting_age_days: 45

# -- Trust filter (optional) --
# Enriches each scanned job with a trust score (0-100) and flags.
# Jobs are NEVER dropped — only annotated. The trust score helps you
# prioritize which postings to evaluate first.
#
# Heuristics (V1):
#   1. URL structure validation (malformed or non-http URLs)
#   2. Missing application URL
#   3. Suspicious domain detection (link shorteners, form builders)
#   4. Company ↔ domain mismatch (with ATS allowlist)
#
# If this block is absent or enabled: false, all jobs get score 100.
#
# trust_filter:
#   enabled: true
#   suspicious_domains:
#     - "bit.ly"
#     - "tinyurl.com"
#     - "t.co"
#     - "forms.gle"
#     - "goo.gl"
#     - "shorturl.at"
#   # Override the built-in ATS allowlist (greenhouse, lever, workday, etc.).
#   # URLs hosted on these domains skip the company↔domain mismatch check.
#   # ats_allowlist:
#   #   - "greenhouse.io"
#   #   - "ashbyhq.com"
#   #   - "lever.co"
#   #   - "custom-ats.example.com"

# -- Content filter (optional) --
# Filters on the job DESCRIPTION text, not just the title — useful when
# companies post generic titles ("Software Engineer") with very different
# stacks. Case-insensitive substring match.
# Semantics:
#   - A job with no description always passes (see provider support below)
#   - Any `negative` keyword present → reject
#   - `positive` empty → pass (after clearing negatives)
#   - `positive` non-empty → at least one keyword must be present
#   - If this entire block is absent, descriptions are not filtered (default)
#
# Provider support: the scanner is zero-token and only reads descriptions that
# a provider already returns in its list payload — it never fetches each job
# page. Currently Lever (`descriptionPlain`) supplies descriptions; jobs from
# providers that don't always pass this filter (they're never silently dropped).
#
# Example: keep roles that mention Rust/Go, drop legacy PHP/WordPress shops
# content_filter:
#   positive:
#     - "rust"
#     - "golang"
#   negative:
#     - "php"
#     - "wordpress"
#
# by_title_keyword (optional): scope a stricter positive/negative pair to only
# the jobs matched via a specific title_filter.positive keyword, instead of
# applying it to every scanned job. Useful when a title keyword is broad
# enough to need extra confirmation from the description (e.g. "AI Engineer"
# roles where you want the JD to actually mention a concrete AI tool) without
# that same requirement leaking onto unrelated categories (Instructional
# Designer, HR Coordinator, ...). Jobs matched via a keyword with no entry
# here fall back to the global positive/negative pair above.
# content_filter:
#   by_title_keyword:
#     "AI Engineer":
#       positive:
#         - "gpt"
#         - "llm"
#         - "claude"

# -- Title filter --
# The scanner uses these keywords to decide if a title is relevant.
# At least 1 positive must match AND 0 negatives must match (case-insensitive).
#
# skip_tiers: [intern, entry]  # optional: skip listings by seniority tier (note: unrecognized/plain titles fall back to 'mid')

title_filter:
  positive:
    # [CUSTOMIZE] Add keywords matching YOUR target roles
    # -- AI/ML roles --
    - "AI"
    - "ML"
    - "LLM"
    - "Agent"
    - "Agentic"
    - "GenAI"
    - "Generative AI"
    - "NLP"
    - "LLMOps"
    - "MLOps"
    - "Voice AI"
    - "Conversational AI"
    - "Speech"
    # -- German AI/ML keywords (DACH market) --
    - "KI"
    - "Künstliche Intelligenz"
    - "KI Engineer"
    - "KI Trainer"
    - "Dozent"
    - "Weiterbildung"
    # -- Engineering roles --
    - "Platform Engineer"
    - "Solutions Architect"
    - "Solutions Engineer"
    - "Forward Deployed"
    - "Deployed Engineer"
    - "Customer Engineer"
    - "Integration Engineer"
    # -- Product roles --
    - "Product Manager"
    - "Technical PM"
    # -- Automation roles --
    - "Automation"
    - "Hyperautomation"
    - "Low-Code"
    - "No-Code"
    - "GTM Engineer"
    - "RevOps"
    - "Business Systems"
    - "Internal Tools"
    - "Transformation"
  negative:
    # [CUSTOMIZE] Add keywords for roles you want to exclude
    - "Junior"
    - "Intern"
    - ".NET"
    - "Java "
    - "iOS"
    - "Android"
    - "PHP"
    - "Ruby"
    - "Embedded"
    - "Firmware"
    - "FPGA"
    - "ASIC"
    - "Blockchain"
    - "Web3"
    - "Crypto"
    - "Salesforce Admin"
    - "SAP "
    - "Oracle EBS"
    - "Mainframe"
    - "COBOL"
  seniority_boost:
    # These prefixes add relevance but are not required
    - "Senior"
    - "Staff"
    - "Principal"
    - "Lead"
    - "Head"
    - "Director"

# -- Search queries --
# Each query triggers a WebSearch. Use site: to filter by portal.
# The scanner extracts title, URL, and company from results.

search_queries:
  # -- Ashby --
  - name: Ashby — AI PM
    query: 'site:jobs.ashbyhq.com "AI Product Manager" OR "Senior Product Manager AI" remote'
    enabled: true

  - name: Ashby — Solutions Architect
    query: 'site:jobs.ashbyhq.com "Solutions Architect" AI OR automation remote'
    enabled: true

  - name: Ashby — AI Engineer
    query: 'site:jobs.ashbyhq.com "AI Engineer" OR "LLM Engineer" OR "Forward Deployed" remote'
    enabled: true

  - name: Ashby — Agentic
    query: 'site:jobs.ashbyhq.com "Agentic" OR "AI Agent" OR "Automation Architect" remote'
    enabled: true

  # -- Greenhouse --
  - name: Greenhouse — AI PM
    query: 'site:boards.greenhouse.io OR site:job-boards.greenhouse.io "AI Product Manager" OR "Senior Product Manager" AI remote'
    enabled: true

  - name: Greenhouse — SA & FDE
    query: 'site:boards.greenhouse.io OR site:job-boards.greenhouse.io "Solutions Architect" OR "Forward Deployed" AI remote'
    enabled: true

  - name: Greenhouse — AI Engineer
    query: 'site:boards.greenhouse.io OR site:job-boards.greenhouse.io "AI Engineer" OR "LLM" OR "Agentic" remote'
    enabled: true

  # -- Workable --
  - name: Workable — AI Roles
    query: 'site:apply.workable.com "AI" "Product Manager" OR "Solutions Architect" OR "AI Engineer" remote'
    enabled: true

  # -- No-Code / Low-Code / Automation --
  - name: Ashby — No-Code & Automation
    query: 'site:jobs.ashbyhq.com "no-code" OR "low-code" OR "automation engineer" OR "n8n" OR "Make" OR "Airtable" remote senior'
    enabled: true

  - name: Greenhouse — No-Code & Automation
    query: 'site:job-boards.greenhouse.io "no-code" OR "low-code" OR "automation engineer" OR "business systems" senior remote'
    enabled: true

  - name: GTM Engineer — All portals
    query: '"GTM Engineer" OR "RevOps Engineer" OR "Growth Engineer" automation Airtable OR Make OR Clay remote'
    enabled: true

  # -- Voice AI & Conversational AI --
  - name: Voice AI — FDE & SA
    query: 'site:job-boards.greenhouse.io OR site:jobs.ashbyhq.com "Voice AI" OR "Conversational AI" OR "Speech" "Solutions" OR "Forward Deployed" OR "Customer Engineer"'
    enabled: true

  - name: Voice AI — AI Engineer
    query: 'site:job-boards.greenhouse.io OR site:jobs.ashbyhq.com "Voice" OR "Speech" "AI Engineer" OR "Deployment" OR "Integration"'
    enabled: true

  # -- Contact Center AI --
  - name: Contact Center AI — SA & SE
    query: '"contact center" OR "customer service" "AI" "Solutions Architect" OR "Solutions Engineer" OR "Forward Deployed" remote'
    enabled: true

  # -- FDE specific (cross-portal) --
  - name: FDE & Deployed Engineer — All portals
    query: '"Forward Deployed Engineer" OR "Deployed Engineer" OR "Deployment Engineer" AI site:job-boards.greenhouse.io OR site:jobs.ashbyhq.com OR site:jobs.lever.co'
    enabled: true

  # -- Remote-friendly Europe boards --

  - name: Remotive — AI/DevRel/SA/FDE
    query: 'site:remotive.com "AI Engineer" OR "Developer Advocate" OR "Solutions Architect" OR "Forward Deployed" OR "DevRel" Europe'
    enabled: true

  - name: WeWorkRemotely — AI & DX
    query: 'site:weworkremotely.com "AI" OR "DevRel" OR "Solutions Engineer" OR "Forward Deployed" OR "Founding Engineer"'
    enabled: true

  - name: Working Nomads — AI/DevRel/SA
    query: 'site:workingnomads.com "AI Engineer" OR "Developer Advocate" OR "Solutions Architect" OR "Forward Deployed"'
    enabled: true

  - name: EU Remote Jobs — AI/DevRel/SA
    query: 'site:euremotejobs.com "AI" OR "DevRel" OR "Solutions Architect" OR "Forward Deployed" OR "Founding Engineer"'
    enabled: true

  - name: RemoteOK — AI/DevRel/SA
    query: 'site:remoteok.com "AI Engineer" OR "Developer Advocate" OR "Solutions Architect" OR "Forward Deployed" OR "Founding Engineer"'
    enabled: true

  # -- Specialized AI / Dev-Tools / FDE boards --

  - name: ai-jobs.net — DevRel/SA/FDE
    query: 'site:ai-jobs.net "Developer Advocate" OR "Solutions Architect" OR "Forward Deployed" OR "DevRel" remote Europe'
    enabled: true

  - name: ai-jobs.net — Founding/Applied AI
    query: 'site:ai-jobs.net "Founding Engineer" OR "Applied AI Engineer" OR "Founding" remote'
    enabled: true

  - name: fwddeploy.com — Forward Deployed roles
    query: 'site:fwddeploy.com "Forward Deployed" OR "Deployed Engineer" remote Europe OR EMEA'
    enabled: true

  - name: YC Jobs — DevRel/SA/FDE/Founding
    query: 'site:ycombinator.com/jobs OR site:workatastartup.com "Forward Deployed" OR "Developer Advocate" OR "Founding Engineer" OR "Solutions Architect" remote'
    enabled: true

  - name: YC Jobs — Applied AI
    query: 'site:workatastartup.com "Applied AI" OR "AI Engineer" OR "AI Agent" remote Europe'
    enabled: true

  # -- Spanish market specialized (high-value for Madrid/Barcelona-based candidates) --

  - name: Manfred — Tech Senior España
    query: 'site:getmanfred.com/en/job-offers "Solutions Architect" OR "DevRel" OR "Forward Deployed" OR "AI Engineer" OR "Founding"'
    enabled: true

  - name: Manfred — Remote AI/Senior
    query: 'site:getmanfred.com "remote" "AI" OR "Senior" OR "Staff" OR "Lead"'
    enabled: true

  - name: Tecnoempleo — España IT
    query: 'site:tecnoempleo.com "AI Engineer" OR "Solutions Architect" OR "DevRel" OR "Forward Deployed" OR "Founding Engineer"'
    enabled: true

  - name: Tecnoempleo — Remoto Senior
    query: 'site:tecnoempleo.com "remoto" OR "remote" Senior "AI" OR "Architect" OR "Lead"'
    enabled: true

  - name: JobFluent — Spain startups
    query: 'site:jobfluent.com Madrid OR Barcelona "AI" OR "Engineer" OR "Solutions" OR "DevRel"'
    enabled: true

  - name: Welcome to the Jungle — Spain tech
    query: 'site:welcometothejungle.com Spain OR Madrid OR Barcelona "AI Engineer" OR "Solutions Architect" OR "DevRel" OR "Forward Deployed"'
    enabled: true

  # -- DevRel-specific boards --

  - name: DevRelX Jobs
    query: 'site:devrelx.com/jobs remote OR Europe OR EMEA'
    enabled: true

  - name: TrueUp — DevRel roles
    query: 'site:trueup.io/devrel OR site:trueup.io "Developer Advocate" OR "DevRel" remote Europe'
    enabled: true

  - name: Remote Rocketship — DevRel
    query: 'site:remoterocketship.com/jobs/developer-relations OR site:remoterocketship.com "DevRel" remote'
    enabled: true

  - name: DevRel Job board
    query: 'site:devreljob.com remote OR Europe OR EMEA'
    enabled: true

  # -- European / global aggregators --

  - name: Himalayas — Remote AI/Dev-tools
    query: 'site:himalayas.app "AI Engineer" OR "Solutions Architect" OR "Developer Advocate" OR "Forward Deployed" OR "Founding Engineer" Europe'
    enabled: true

  - name: Jooble — España Tech Senior
    query: 'site:jooble.org "AI Engineer" OR "Solutions Architect" OR "DevRel" Madrid OR remote España'
    enabled: true

  - name: EU Data Jobs — AI/ML Europe
    query: 'site:eudatajobs.com "AI" OR "ML" OR "Data" "Engineer" OR "Architect" remote Europe'
    enabled: true

  - name: HN Who's Hiring — Remote AI
    query: 'site:news.ycombinator.com OR site:hnrss.org "AI Engineer" OR "Forward Deployed" OR "DevRel" OR "Founding Engineer" remote'
    enabled: true

  # -- Türkiye / Turkey job portals --
  # WebSearch queries for Turkish job boards (no public ATS APIs).
  # [CUSTOMIZE] Replace keywords with your target roles in Turkish.

  - name: Kariyer.net — Backend / Software Engineer
    query: 'site:kariyer.net "Backend Developer" OR "Software Engineer" OR "Yazılım Geliştirici" OR "Backend Geliştirici" Ankara OR Istanbul OR uzaktan'
    enabled: false

  - name: Kariyer.net — Frontend Engineer
    query: 'site:kariyer.net "Frontend Developer" OR "React Developer" OR "Frontend Geliştirici" Ankara OR Istanbul OR uzaktan'
    enabled: false

  - name: Yenibiris.com — Software Engineer
    query: 'site:yenibiris.com "Software Engineer" OR "Backend Developer" OR "Frontend Developer" OR "Yazılım Geliştirici"'
    enabled: false

  - name: Secretcv.com — Tech Roles
    query: 'site:secretcv.com "Software Engineer" OR "Backend" OR "Frontend" OR "Full Stack"'
    enabled: false

  - name: İşin Olsun — Startup / Tech
    query: 'site:isinolsun.com "Yazılım" OR "Software" OR "Backend" OR "Frontend" OR "Full Stack"'
    enabled: false

  - name: LinkedIn — Turkey Tech Roles
    query: 'site:linkedin.com/jobs "Software Engineer" OR "Backend Engineer" OR "Frontend Engineer" Turkey OR Türkiye OR Ankara OR Istanbul'
    enabled: false

# -- Tracked companies --
# Companies whose career pages are checked directly.
# scan_method: playwright (default), websearch, local_parser
# For Greenhouse companies, add api: field for faster structured JSON access.
#
# Provider routing (scan.mjs):
#   By default each entry is auto-detected from `careers_url` against the
#   loaded providers in providers/*.mjs (local-parser, ashby, greenhouse, lever).
#   local-parser runs first when parser.command + parser.script are set.
#
#   Optional fields on a tracked_companies entry:
#     provider: <id>     — force a specific provider, skipping detect().
#     transport: http    — reserved for future transports. Defaults to http.
#
# For SSR/static career pages, add parser.command + parser.script. The command
# can be python3, node, or another local executable. The script must print JSON
# jobs to stdout: [{ "title": "...", "url": "...", "location": "..." }]
#
# Generic local parser examples. Most parser scripts are company-specific and
# already know their target careers URL. `args` are optional and can help the
# parser author (reuse across companies, debug flags, JSON snapshots, etc.).
#
# JavaScript parser:
# - name: Example JS Company
#   careers_url: https://example.com/careers
#   scan_method: local_parser
#   parser:
#     command: node
#     script: scripts/parsers/example-js-company-jobs.js
#     format: jobs-json-v1
#   enabled: true
#
# Python parser:
# - name: Example Python Company
#   careers_url: https://example.org/jobs
#   scan_method: local_parser
#   parser:
#     command: python3
#     script: scripts/parsers/example_python_company_jobs.py
#     format: jobs-json-v1
#   enabled: true

# ── Provider auto-detection ───────────────────────────────────────
# scan.mjs auto-loads everything in providers/*.mjs and tries each
# provider's detect() in order. URL patterns recognized:
#
#   greenhouse      job-boards(.eu)?.greenhouse.io/<slug>  (or api: field)
#   ashby           jobs.ashbyhq.com/<slug>
#   lever           jobs.(eu.)?lever.co/<slug>
#   workable        apply.workable.com/<slug>
#   smartrecruiters (careers|jobs).smartrecruiters.com/<slug>
#   recruitee       <slug>.recruitee.com
#   bamboohr        <tenant>.bamboohr.com               (or api: field)
#   breezy          <tenant>.breezy.hr                  (or api: field)
#   solidjobs       solid.jobs/public-api/offers/<division>
#   comeet          api: https://www.comeet.co/careers-api/2.0/company/<uid>/positions?token=<token>
#   personio        <slug>.jobs.personio.(de|com)
#   pinpoint        <slug>.pinpointhq.com
#   rippling        ats.rippling.com/<slug>/jobs
#   jibeapply       <slug>.jibeapply.com/jobs  (or provider: jibeapply + api: for a branded host's own /jobs path)
#   tencent         careers.tencent.com  (zh-CN board; each keywords: entry queried server-side, results deduped)
#   workday         <tenant>.wd<N>.myworkdayjobs.com[/<locale>]/<board>
#   avature         <tenant>.avature.net  (branded host → provider: avature + api: the /careers/SearchJobs URL)
#   amazon          amazon.jobs           (global board; provider: amazon + an amazon: filter block, see below)
#   successfactors  *.successfactors.(eu|com) / *.jobs2web.com  (SAP SF Recruiting Marketing / RMK boards;
#                   branded hosts like jobs.<company>.com need provider: successfactors + api: <board origin>)
#
# When the public careers URL is a branded custom domain (e.g.
# careers.adyen.com), set `provider: smartrecruiters` explicitly to
# bypass detect(). The `provider:` field wins over auto-detection.
#
# Some providers are job boards / aggregators with no per-company careers
# URL to detect, so they must be selected with an explicit `provider:`
# field: 4dayweek, arbeitnow, arbeitsagentur, getonbrd, glints, himalayas, ibm, jobicy,
# jobstreet, justjoin, landingjobs, nofluffjobs, remoteok, remotive, thehub,
# weworkremotely, workingnomads. Examples for every built-in provider are below.

# ── Built-in provider examples ────────────────────────────────────
# Commented config for each provider in providers/. Copy a stanza into
# tracked_companies, uncomment, and adjust. Auto-detected providers only
# need careers_url; the rest need provider: plus any fields shown.
#
# -- Auto-detected company ATS (careers_url is enough) --
#
# - name: Example BambooHR Co
#   careers_url: https://exampleco.bamboohr.com/careers   # any *.bamboohr.com host
#   enabled: true
#
# - name: Example Breezy Co
#   careers_url: https://exampleco.breezy.hr              # any *.breezy.hr host
#   enabled: true
#
# - name: Example Pinpoint Co
#   careers_url: https://exampleco.pinpointhq.com         # any *.pinpointhq.com host
#   enabled: true
#
# - name: Example Rippling Co
#   careers_url: https://ats.rippling.com/exampleco/jobs  # any ats.rippling.com/<slug>
#   enabled: true
#
# - name: Example JibeApply Co
#   careers_url: https://exampleco.jibeapply.com/jobs     # any *.jibeapply.com/jobs host
#   max_pages: 50          # optional page cap (10 jobs/page, default 50, max 500) —
#                          # raise it for a tenant with 500+ open postings (a warning
#                          # is logged if the cap truncates real results)
#   enabled: true
#
# - name: 腾讯 (Tencent)
#   careers_url: https://careers.tencent.com/search.html  # any https://careers.tencent.com URL
#   keywords: ["AI", "大模型"]   # each keyword is a separate server-side query, results deduped;
#                                # omit keywords to pull the whole board
#   max_pages: 20          # optional per-keyword page cap (100 posts/page, default 20)
#   enabled: true
#
# - name: Example Workday Co
#   careers_url: https://exampleco.wd5.myworkdayjobs.com/en-US/exampleco-careers  # any <tenant>.wd<N>.myworkdayjobs.com
#   max_pages: 100         # optional page cap (20 jobs/page, default 100, max 1500) —
#                          # raise it for a tenant with 2000+ open postings (a warning
#                          # is logged if the cap truncates real results)
#   enabled: true
#
# - name: Example Avature Co
#   # Avature career site. Auto-detects any *.avature.net host; for a branded
#   # host that proxies Avature, set provider: avature + api: the SearchJobs URL.
#   careers_url: https://exampleco.avature.net/careers/SearchJobs
#   max_pages: 50          # optional page cap (only 6 jobs/page — Avature's hard
#                          # limit — so ~300 postings by default, max 200 pages)
#   enabled: true
#
# - name: Example SuccessFactors Co
#   # SAP SuccessFactors Recruiting Marketing (RMK, ex-jobs2web) branded board —
#   # what many large industrials run at their own domain (jobs.<company>.com).
#   # Branded hosts carry no "successfactors" string, so set provider: explicitly
#   # and point api: at the board origin (host only, no path). The provider pulls
#   # the public /tile-search-results fragment (zero-token, no auth).
#   careers_url: https://www.exampleco.com/careers   # public/branded careers page
#   provider: successfactors
#   api: https://jobs.exampleco.com                  # RMK board origin
#   enabled: true
#
# - name: SolidJobs — Engineering
#   # division path is required; one of:
#   # it, engineering, marketing, sales, hr, logistics, finances, other
#   careers_url: https://solid.jobs/public-api/offers/engineering
#   enabled: true
#
# -- Job boards / aggregators (no careers_url; set provider: explicitly) --
#
# - name: Amazon / AWS                    # global amazon.jobs board
#   provider: amazon
#   amazon:                # keys pass straight through to the search.json API
#     normalized_country_code: [DEU]   # array → facet filter (…[]=DEU); REQUIRED-ish,
#                                       # the board is 100k+ postings otherwise
#     base_query: machine learning      # optional keyword filter
#     # loc_query: Berlin               # optional free-text location (fuzzy, not a hard filter)
#   enabled: true
#
# - name: 4 Day Week                      # 4-day-week / reduced-hours roles
#   provider: 4dayweek
#   max_pages: 3          # optional page cap (25 jobs/page, default 3, max 50)
#   enabled: true
#
# - name: The Hub                         # Nordic/EU startup board
#   provider: thehub
#   max_pages: 3          # optional page cap (15 jobs/page, default 3, max 67)
#   enabled: true
#
# - name: Arbeitnow                       # EU/DACH-heavy board, newest-first
#   provider: arbeitnow
#   max_pages: 3          # optional page cap (100 jobs/page, default 3, max 50)
#   enabled: true
#
# - name: Arbeitsagentur — AI Roles      # German federal jobs API
#   provider: arbeitsagentur
#   arbeitsagentur:
#     keywords: ["AI Engineer", "Machine Learning"]  # required (>=1)
#     wo: Berlin          # optional anchor city ('' / omit = all of Germany)
#     umkreis: 50         # optional km radius around wo (0-1000)
#     days: 30            # optional recency window (1-1000)
#     size: 100           # optional results per keyword (1-100)
#   enabled: true
#
# - name: Glints — Indonesia             # SE-Asia aggregator (SG/ID/MY/VN)
#   provider: glints
#   searchKeywords: "AI Engineer"
#   countryCode: ID       # optional 2-letter code (default ID)
#   enabled: true
#
# - name: Jobstreet — Indonesia          # SEEK aggregator
#   provider: jobstreet
#   searchKeywords: "Software Engineer"
#   siteKey: ID-Main      # SEEK site key (ID-Main, SG-Main, MY-Main)
#   searchLocation: ""    # optional location filter
#   enabled: true
#
# - name: IBM — Germany SWE              # IBM global careers
#   provider: ibm
#   ibm:
#     country: Germany                      # optional facet filter
#     categories: ["Software Engineering"]  # optional facet filter
#   enabled: true
#
# - name: Dassault Systèmes              # single-company; Exalead card_search_api
#   careers_url: https://www.3ds.com/careers/jobs   # auto-detects on *.3ds.com
#   provider: dassault                    # optional — 3ds.com host also auto-detects
#   enabled: true                         # ~445 English postings (zero-token XML)
#
# - name: Deutsche Bahn                   # single-company; db.jobs Avature front
#   careers_url: https://db.jobs          # auto-detects; SSR search, paginated over HTTP
#   provider: deutschebahn                # optional — db.jobs host also auto-detects
#   enabled: true
#
# - name: Heckler & Koch                  # single-company; SSR Nuxt vacancy list
#   careers_url: https://www.heckler-koch.com/de/Karriere/Stellenangebote
#   provider: hecklerkoch                 # optional — heckler-koch.com host also auto-detects
#   enabled: true
#
# - name: Rheinmetall                     # single-company; SSR Nuxt vacancy list
#   careers_url: https://www.rheinmetall.com/en/career/vacancies
#   provider: rheinmetall                 # optional — rheinmetall.com host also auto-detects
#   enabled: true
#
# - name: OHB                             # Cornerstone OnDemand (csod) branded site
#   careers_url: https://career-ohb.csod.com/ux/ats/careersite/4/home?c=career-ohb
#   provider: csod                        # anonymous-JWT bootstrap, then public search API
#   enabled: true
#
# - name: RENK                            # softgarden hosted job widget
#   careers_url: https://renk-group.softgarden.io/de/widgets/jobs
#   provider: softgarden
#   enabled: true
#
# - name: Mercedes-Benz                   # BeeSite (milch & zucker) search backend
#   careers_url: https://mercedes-benz.app.beesite.de   # the *.app.beesite.de host, not the branded front
#   provider: beesite
#   enabled: true
#
# -- Remote-job feeds (no careers_url, no extra fields) --
# title_filter / location_filter defined above still apply.
#
# - name: Himalayas
#   provider: himalayas
# - name: Landing.jobs                    # tech board, Europe-focused
#   provider: landingjobs
#   enabled: true
#
# - name: RemoteOK
#   provider: remoteok
#   enabled: true
#
# - name: Remotive
#   provider: remotive
#   enabled: true
#
# - name: Get on Board
#   provider: getonbrd
#   enabled: true
#
# - name: NoDesk
#   provider: nodesk
#   enabled: true
#
# - name: We Work Remotely
#   provider: weworkremotely
#   enabled: true
#
# - name: Working Nomads
#   provider: workingnomads
#   enabled: true
#
# - name: Jobicy
#   provider: jobicy
#   enabled: true
#
# - name: EchoJobs                        # tech jobs aggregated from company ATS boards
#   provider: echojobs
#   max_pages: 3                          # 100 jobs/page; raise for deeper coverage
#   enabled: true
#
# - name: Hacker News "Who is hiring?"    # monthly "Ask HN: Who is hiring?" thread
#   provider: hackernews
#   enabled: true
#
# - name: Jobspresso                      # curated remote roles (public WordPress feed)
#   provider: jobspresso
#   enabled: true
#
# - name: The Muse                        # board-wide public jobs API
#   provider: themuse
#   enabled: true
#
# - name: LaraJobs                        # Laravel / PHP jobs (public RSS feed)
#   provider: larajobs
#   enabled: true
#
# Comeet needs the full careers-api URL (it carries a per-tenant token that a
# branded careers page does not expose), so set it via the api: field — e.g.:
#
#   - name: Spark Hire
#     careers_url: https://www.comeet.com/jobs/spark-hire/30.005
#     api: https://www.comeet.co/careers-api/2.0/company/30.005/positions?token=YOUR_TOKEN
#     enabled: true
#
# Find <uid> and <token> in Comeet under Settings → Sourcing → Careers Website →
# Integrated website → Careers API.
#
# Personio (common across DACH/EU) auto-detects from its careers host and
# pulls the public no-auth XML feed at <slug>.jobs.personio.de/xml — e.g.:
#
#   - name: Example GmbH
#     careers_url: https://example-gmbh.jobs.personio.de
#     enabled: true
#
# Teamtailor auto-detects from a <slug>.teamtailor.com careers host and reads
# the public no-auth /jobs.rss feed — e.g.:
#
#   - name: Example Co
#     careers_url: https://example.teamtailor.com
#     enabled: true
#
#   # If the tenant fronts a branded careers domain, set provider: teamtailor
#   # explicitly and point careers_url at that host (it serves the same feed):
#   - name: Example Co (branded)
#     careers_url: https://careers.example.com
#     provider: teamtailor
#     enabled: true
#
# Some JibeApply sites run on a branded custom domain rather than
# *.jibeapply.com (iCIMS acquired Jibe in 2019, and tenants can front the
# same API on their own domain). Set provider: jibeapply and point api: at
# the branded host's /api/jobs endpoint — e.g.:
#
#   - name: Booking.com
#     careers_url: https://jobs.booking.com/booking/jobs
#     api: https://jobs.booking.com/api/jobs
#     provider: jibeapply
#     enabled: true

# -- Company aliases (optional) --
# Map alternate company names to a single canonical label so the scanner's
# company+role dedup matches your tracker. The scanner keys dedup on the name a
# provider returns — usually the ATS org (e.g. Greenhouse "Intercom") — which
# may differ from the brand you track (e.g. "Fin"). Without an alias, the same
# role is re-scanned and re-evaluated every run because "intercom::role" never
# matches "fin::role" in data/applications.md.
#
# Format: `CanonicalName: [alias, ...]`. Every alias AND the canonical name
# resolve to the canonical label (case-insensitive). Companies with no entry are
# unaffected. Combined with title normalization, this suppresses URL-new
# duplicates caused by requisition URL changes and per-location title splits
# ("Engineer (Berlin)" vs "Engineer"). It does not classify repost patterns or
# produce `postingChurn` evidence; detect-reposts.mjs and company-history.mjs own
# that post-tracking layer.
#
# company_aliases:
#   Fin: [Intercom]

tracked_companies:

  # -- AI Labs & LLM providers --

  - name: Anthropic
    careers_url: https://job-boards.greenhouse.io/anthropic
    api: https://boards-api.greenhouse.io/v1/boards/anthropic/jobs
    enabled: true

  - name: OpenAI
    careers_url: https://openai.com/careers
    scan_method: websearch
    scan_query: 'site:openai.com/careers "Solutions" OR "Forward Deployed" OR "AI Engineer"'
    enabled: true

  # -- Voice AI & Conversational AI --

  - name: PolyAI
    careers_url: https://job-boards.eu.greenhouse.io/polyai
    api: https://boards-api.greenhouse.io/v1/boards/polyai/jobs
    notes: "UK. Voice AI for enterprise contact centers."
    enabled: true

  - name: Parloa
    careers_url: https://job-boards.eu.greenhouse.io/parloa
    api: https://boards-api.greenhouse.io/v1/boards/parloa/jobs
    notes: "Berlin EMEA. Voice AI enterprise."
    enabled: true

  - name: Intercom
    careers_url: https://job-boards.greenhouse.io/intercom
    api: https://boards-api.greenhouse.io/v1/boards/intercom/jobs
    notes: "Dublin EMEA. Fin AI agent."
    enabled: true

  - name: Hume AI
    careers_url: https://job-boards.greenhouse.io/humeai
    api: https://boards-api.greenhouse.io/v1/boards/humeai/jobs
    notes: "NYC. Empathic voice AI."
    enabled: true

  - name: ElevenLabs
    careers_url: https://jobs.ashbyhq.com/elevenlabs
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/elevenlabs'
    notes: "Voice AI TTS leader."
    enabled: true

  - name: Deepgram
    careers_url: https://jobs.ashbyhq.com/deepgram
    scan_method: websearch
    scan_query: 'site:deepgram.com/careers OR "deepgram" "Solutions" OR "AI Engineer" site:jobs.ashbyhq.com'
    notes: "STT/TTS APIs."
    enabled: true

  - name: Vapi
    careers_url: https://jobs.ashbyhq.com/vapi
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/vapi'
    notes: "Voice AI infrastructure."
    enabled: true

  - name: Bland AI
    careers_url: https://jobs.ashbyhq.com/bland
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/bland'
    notes: "Voice phone agents. $65M Series B."
    enabled: true

  # -- AI-native platforms (FDE/SA teams) --

  - name: Retool
    careers_url: https://retool.com/careers
    scan_method: websearch
    scan_query: 'site:retool.com/careers "Deployed" OR "AI" OR "Solutions"'
    notes: "London. Popularized Deployed Engineer role."
    enabled: true

  - name: Airtable
    careers_url: https://job-boards.greenhouse.io/airtable
    api: https://boards-api.greenhouse.io/v1/boards/airtable/jobs
    notes: "No-code + AI platform."
    enabled: true

  - name: Vercel
    careers_url: https://job-boards.greenhouse.io/vercel
    api: https://boards-api.greenhouse.io/v1/boards/vercel/jobs
    notes: "AI SDK, v0.dev. Frontend AI tooling."
    enabled: true

  - name: Temporal
    careers_url: https://job-boards.greenhouse.io/temporal
    api: https://boards-api.greenhouse.io/v1/boards/temporal/jobs
    notes: "Workflow orchestration."
    enabled: true

  - name: Arize AI
    careers_url: https://job-boards.greenhouse.io/arizeai
    api: https://boards-api.greenhouse.io/v1/boards/arizeai/jobs
    notes: "LLMOps / AI observability."
    enabled: true

  - name: RunPod
    careers_url: https://job-boards.greenhouse.io/runpod
    api: https://boards-api.greenhouse.io/v1/boards/runpod/jobs
    notes: "GPU cloud for AI."
    enabled: true

  - name: Weights & Biases (CoreWeave)
    careers_url: https://job-boards.greenhouse.io/coreweave
    api: https://boards-api.greenhouse.io/v1/boards/coreweave/jobs
    notes: "Acquired by CoreWeave (announced March 2025, closed May 2025). W&B roles now live on CoreWeave's Greenhouse board. Old Lever URL (jobs.lever.co/wandb) returns 404."
    enabled: true

  - name: Glean
    careers_url: https://job-boards.greenhouse.io/gleanwork
    api: https://boards-api.greenhouse.io/v1/boards/gleanwork/jobs
    notes: "Enterprise AI search."
    enabled: true

  # -- Contact Center AI & CX --

  - name: Ada
    careers_url: https://job-boards.greenhouse.io/ada
    scan_method: websearch
    scan_query: '"ada.cx" OR "ada" "AI" "Solutions" OR "Engineer" site:jobs.ashbyhq.com OR site:job-boards.greenhouse.io'
    notes: "Toronto + remote. AI customer service."
    enabled: true

  - name: LivePerson
    careers_url: https://liveperson.com/company/careers
    scan_method: websearch
    scan_query: 'site:liveperson.com/careers OR "liveperson" "Solutions" OR "AI" site:job-boards.greenhouse.io'
    notes: "Remote EMEA. Conversational AI enterprise."
    enabled: true

  - name: Sierra
    careers_url: https://jobs.ashbyhq.com/sierra
    scan_method: websearch
    scan_query: '"sierra.ai" "AI" "Forward Deployed" OR "Applied" OR "Solutions" site:jobs.ashbyhq.com'
    notes: "Bret Taylor (ex-CEO Salesforce). AI customer agents."
    enabled: true

  - name: Decagon
    careers_url: https://jobs.ashbyhq.com/decagon
    scan_method: websearch
    scan_query: '"decagon" "AI" "Forward Deployed" OR "Solutions" site:jobs.ashbyhq.com'
    notes: "AI customer support agents."
    enabled: true

  # -- Enterprise comms & contact center --

  - name: Talkdesk
    careers_url: https://www.talkdesk.com/careers
    scan_method: websearch
    scan_query: 'site:talkdesk.com/careers "AI" OR "Solutions Architect" OR "AI Engineer"'
    notes: "Lisbon. Contact center AI. EMEA friendly."
    enabled: true

  - name: Twilio
    careers_url: https://www.twilio.com/en-us/company/jobs
    scan_method: websearch
    scan_query: 'site:twilio.com/company/jobs "AI" OR "Voice" "Solutions" OR "Architect"'
    notes: "Voice/messaging infrastructure."
    enabled: true

  - name: Dialpad
    careers_url: https://www.dialpad.com/careers
    scan_method: websearch
    scan_query: '"dialpad" "AI" "Solutions" OR "Engineer" site:job-boards.greenhouse.io OR site:jobs.ashbyhq.com'
    notes: "Voice AI for business comms."
    enabled: true

  - name: Gong
    careers_url: https://www.gong.io/careers
    scan_method: websearch
    scan_query: 'site:gong.io/careers "AI" OR "Solutions" OR "Architect"'
    notes: "Revenue intelligence with voice AI."
    enabled: true

  - name: Genesys
    careers_url: https://www.genesys.com/careers
    scan_method: websearch
    scan_query: 'site:genesys.com/careers "AI" "Solutions Architect" OR "AI Engineer"'
    notes: "Contact center cloud + AI."
    enabled: true

  - name: Salesforce
    careers_url: https://careers.salesforce.com
    scan_method: websearch
    scan_query: 'site:careers.salesforce.com "Forward Deployed" OR "AI Engineer" OR "Agentforce"'
    notes: "Agentforce = AI agents platform."
    enabled: true

  # -- AI infra & LLMOps --

  - name: Langfuse
    careers_url: https://langfuse.com/careers
    scan_method: websearch
    scan_query: '"langfuse" careers OR jobs "engineer" OR "solutions"'
    notes: "Berlin. LLMOps / observability."
    enabled: true

  - name: Lindy
    careers_url: https://jobs.ashbyhq.com/lindy
    scan_method: websearch
    scan_query: '"lindy.ai" "AI" "Forward Deployed" OR "Solutions" site:jobs.ashbyhq.com'
    notes: "AI agent management platform."
    enabled: true

  - name: Cognigy
    careers_url: https://careers.cognigy.com
    scan_method: websearch
    scan_query: '"cognigy" careers "AI" OR "Solutions" OR "Engineer"'
    notes: "Dusseldorf EMEA. Conversational AI enterprise."
    enabled: true

  - name: Speechmatics
    careers_url: https://job-boards.greenhouse.io/speechmatics
    api: https://boards-api.greenhouse.io/v1/boards/speechmatics/jobs
    notes: "Cambridge UK. Speech recognition platform."
    enabled: true

  # -- No-Code/Low-Code & Automation --

  - name: n8n
    careers_url: https://jobs.ashbyhq.com/n8n
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/n8n'
    notes: "Berlin + remote EU/US. Workflow automation."
    enabled: true

  - name: Zapier
    careers_url: https://jobs.ashbyhq.com/zapier
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/zapier'
    notes: "Remote-first. Automation platform leader."
    enabled: true

  - name: Make.com (Celonis)
    careers_url: https://www.make.com/en/careers
    scan_method: websearch
    scan_query: 'site:make.com/en/careers OR "make.com" "engineer" OR "solutions" site:jobs.ashbyhq.com'
    notes: "Celonis parent company. Automation platform."
    enabled: true

  - name: Boomi
    careers_url: https://job-boards.greenhouse.io/boomilp
    api: https://boards-api.greenhouse.io/v1/boards/boomilp/jobs
    notes: "Integration and automation platform."
    enabled: true

  # -- Ashby-based (Playwright direct) --

  - name: Cohere
    careers_url: https://jobs.ashbyhq.com/cohere
    notes: "AI/LLM provider. Toronto + remote."
    enabled: true

  - name: LangChain
    careers_url: https://jobs.ashbyhq.com/langchain
    notes: "LangChain/LangSmith framework."
    enabled: true

  - name: Pinecone
    careers_url: https://jobs.ashbyhq.com/pinecone
    notes: "Vector database."
    enabled: true

  # -- Lever-based (Playwright direct) --

  - name: Mistral AI
    careers_url: https://jobs.lever.co/mistral
    notes: "Paris. European AI lab."
    enabled: true

  - name: Palantir
    careers_url: https://jobs.lever.co/palantir
    notes: "FDE roles. Mostly US/London."
    enabled: true

  # -- Canada / Vancouver tech & AI --

  - name: Later
    careers_url: https://job-boards.greenhouse.io/later
    api: https://boards-api.greenhouse.io/v1/boards/later/jobs
    notes: "Vancouver/Toronto. Social media and creator platform."
    enabled: true

  - name: Safari AI
    careers_url: https://job-boards.greenhouse.io/safariai
    api: https://boards-api.greenhouse.io/v1/boards/safariai/jobs
    notes: "Miami HQ / Vancouver engineering. Computer vision AI for operations in the physical economy."
    enabled: true

  - name: Hootsuite
    careers_url: https://job-boards.greenhouse.io/hootsuite
    api: https://boards-api.greenhouse.io/v1/boards/hootsuite/jobs
    notes: "Vancouver. Social media management platform."
    enabled: true

  - name: Klue
    careers_url: https://jobs.ashbyhq.com/klue
    notes: "Vancouver. Competitive intelligence SaaS."
    enabled: true

  - name: Glacis AI
    careers_url: https://jobs.ashbyhq.com/glacis-ai
    notes: "Ho Chi Minh City. AI agents for supply chain and logistics workflows."
    enabled: true

  - name: Sanctuary AI
    careers_url: https://jobs.lever.co/sanctuary
    notes: "Vancouver. Humanoid robotics and embodied AI."
    enabled: true

  - name: Semios
    careers_url: https://apply.workable.com/semios/
    scan_method: websearch
    scan_query: 'site:apply.workable.com/semios'
    notes: "Vancouver. Agtech and crop intelligence platform."
    enabled: true

  # -- European tech (EMEA-friendly) --

  - name: Factorial
    careers_url: https://job-boards.greenhouse.io/factorial
    scan_method: websearch
    scan_query: 'site:jobs.greenhouse.io/factorial'
    notes: "Barcelona. HR SaaS unicorn."
    enabled: true

  - name: Attio
    careers_url: https://jobs.ashbyhq.com/attio
    scan_method: websearch
    scan_query: 'site:attio.com/careers OR site:jobs.ashbyhq.com/attio'
    notes: "Remote EU. AI-native CRM. Series B $52M."
    enabled: true

  - name: Tinybird
    careers_url: https://jobs.ashbyhq.com/tinybird
    scan_method: websearch
    scan_query: 'site:tinybird.co/careers OR site:jobs.ashbyhq.com/tinybird'
    notes: "Remote. Real-time data platform."
    enabled: true

  - name: Clarity AI
    careers_url: https://jobs.lever.co/clarity-ai
    scan_method: websearch
    scan_query: 'site:clarity.ai/careers OR site:jobs.lever.co/clarity-ai'
    notes: "Madrid/Remote. Sustainability analytics with AI."
    enabled: true

  - name: Travelperk
    careers_url: https://jobs.ashbyhq.com/travelperk
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/travelperk'
    notes: "Barcelona. Business travel unicorn."
    enabled: true

  # -- DACH & European AI ecosystem --
  # All entries below have been verified against the Greenhouse / Ashby /
  # Lever / Workable APIs or live job boards. Greenhouse boards expose a
  # JSON API at https://boards-api.greenhouse.io/v1/boards/{slug}/jobs which
  # the scanner can hit directly via the api: field.

  # -- DACH AI labs & frontier model providers --

  - name: Aleph Alpha
    careers_url: https://jobs.ashbyhq.com/AlephAlpha
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/AlephAlpha'
    notes: "Heidelberg DE. Sovereign European LLM provider, enterprise & government focus."
    enabled: true

  - name: DeepL
    careers_url: https://jobs.ashbyhq.com/DeepL
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/DeepL'
    notes: "Cologne DE. Translation & language AI. 60+ open roles, EU/DACH heavy."
    enabled: true

  - name: Black Forest Labs
    careers_url: https://job-boards.greenhouse.io/blackforestlabs
    api: https://boards-api.greenhouse.io/v1/boards/blackforestlabs/jobs
    notes: "Freiburg DE / SF. FLUX image models, ex-Stable Diffusion team."
    enabled: true

  - name: Helsing
    careers_url: https://job-boards.greenhouse.io/helsing
    api: https://boards-api.greenhouse.io/v1/boards/helsing/jobs
    notes: "Munich / Berlin / London / Paris. Defence AI unicorn. 100+ roles, ML & FDE."
    enabled: true

  # -- DACH enterprise & SaaS (AI-heavy) --

  - name: Celonis
    careers_url: https://job-boards.greenhouse.io/celonis
    api: https://boards-api.greenhouse.io/v1/boards/celonis/jobs
    notes: "Munich / NYC. Process intelligence, Applied AI Engineer roles, Field CTOs."
    enabled: true

  - name: Contentful
    careers_url: https://job-boards.greenhouse.io/contentful
    api: https://boards-api.greenhouse.io/v1/boards/contentful/jobs
    notes: "Berlin / Denver. Headless CMS, AI content workflows."
    enabled: true

  - name: GetYourGuide
    careers_url: https://job-boards.greenhouse.io/getyourguide
    api: https://boards-api.greenhouse.io/v1/boards/getyourguide/jobs
    notes: "Berlin. Travel marketplace, ML & data platform roles."
    enabled: true

  - name: HelloFresh
    careers_url: https://job-boards.greenhouse.io/hellofresh
    api: https://boards-api.greenhouse.io/v1/boards/hellofresh/jobs
    notes: "Berlin. Meal kit unicorn, large data/ML org."
    enabled: true

  # -- DACH fintech --

  - name: N26
    careers_url: https://job-boards.greenhouse.io/n26
    api: https://boards-api.greenhouse.io/v1/boards/n26/jobs
    notes: "Berlin / Barcelona. Mobile bank, ~50 roles, ML & risk."
    enabled: true

  - name: Trade Republic
    careers_url: https://job-boards.greenhouse.io/traderepublicbank
    api: https://boards-api.greenhouse.io/v1/boards/traderepublicbank/jobs
    notes: "Berlin / London / Paris. Neobroker, ~50 roles, platform & data."
    enabled: true

  - name: SumUp
    careers_url: https://job-boards.greenhouse.io/sumup
    api: https://boards-api.greenhouse.io/v1/boards/sumup/jobs
    notes: "Berlin / London. Payments fintech, EU + LATAM hiring."
    enabled: true

  - name: Qonto
    careers_url: https://jobs.lever.co/qonto
    notes: "Paris / Berlin / Barcelona / Milan. SMB neobank, MLOps for AI Product roles."
    enabled: true

  # -- DACH logistics / mobility --

  - name: Forto
    careers_url: https://jobs.lever.co/forto
    notes: "Berlin DE. Digital freight forwarder, product & data roles, German + English."
    enabled: true

  # -- Switzerland / Austria AI --

  - name: Lakera
    careers_url: https://jobs.ashbyhq.com/lakera.ai
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/lakera.ai'
    notes: "Zurich CH / SF. AI security & guardrails. Has Forward Deployed Engineer roles."
    enabled: true

  - name: Scandit
    careers_url: https://job-boards.greenhouse.io/scandit
    api: https://boards-api.greenhouse.io/v1/boards/scandit/jobs
    notes: "Zurich CH. Computer vision / smart data capture, ML roles."
    enabled: true

  - name: Cradle
    careers_url: https://jobs.ashbyhq.com/cradlebio
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/cradlebio'
    notes: "Zurich CH / Amsterdam. AI-guided protein design, ML researcher roles."
    enabled: true

  # -- France AI ecosystem --

  - name: Hugging Face
    careers_url: https://apply.workable.com/huggingface/
    scan_method: websearch
    scan_query: 'site:apply.workable.com/huggingface'
    notes: "Paris / NYC / Remote. ML hub & open-source models."
    enabled: true

  - name: Photoroom
    careers_url: https://jobs.ashbyhq.com/photoroom
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/photoroom'
    notes: "Paris FR. AI photo editor, Head of CV / Head of ML roles."
    enabled: true

  - name: Pigment
    careers_url: https://jobs.lever.co/pigment
    notes: "Paris FR / NYC / London. AI-powered FP&A planning platform."
    enabled: true

  - name: Diabolocom
    careers_url: https://jobs.eu.lever.co/diabolocom
    notes: "Paris FR. AI-native contact center (CCaaS) platform — Frost & Sullivan 2026 AI CX Leader in Europe; hybrid LLM/SLM architecture powering Agent Assist, Voice Analytics, and automated QA."
    enabled: true

  # -- UK & Ireland AI (EU-friendly hiring) --

  - name: Wayve
    careers_url: https://job-boards.greenhouse.io/wayve
    api: https://boards-api.greenhouse.io/v1/boards/wayve/jobs
    notes: "London UK. Embodied AI for self-driving, ML & robotics roles."
    enabled: true

  - name: Isomorphic Labs
    careers_url: https://job-boards.greenhouse.io/isomorphiclabs
    api: https://boards-api.greenhouse.io/v1/boards/isomorphiclabs/jobs
    notes: "London / Lausanne / Cambridge MA. DeepMind spin-out for drug discovery AI."
    enabled: true

  - name: PhysicsX
    careers_url: https://job-boards.greenhouse.io/physicsx
    api: https://boards-api.greenhouse.io/v1/boards/physicsx/jobs
    notes: "London UK. Physics-informed ML for engineering, ~40 roles."
    enabled: true

  - name: Stability AI
    careers_url: https://job-boards.greenhouse.io/stabilityai
    api: https://boards-api.greenhouse.io/v1/boards/stabilityai/jobs
    notes: "London / SF. Generative AI image/video research lab."
    enabled: true

  - name: Synthesia
    careers_url: https://jobs.ashbyhq.com/synthesia
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/synthesia OR site:job-boards.greenhouse.io/synthesia'
    notes: "London UK. AI video generation for enterprise, $4B valuation."
    enabled: true

  - name: Faculty
    careers_url: https://jobs.ashbyhq.com/faculty
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/faculty'
    notes: "London UK. Applied AI consultancy. 80+ roles, AI Safety / MLE / Delivery."
    enabled: true

  - name: Causaly
    careers_url: https://jobs.ashbyhq.com/causaly
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/causaly'
    notes: "London / Athens. Biomedical knowledge graph + AI."
    enabled: true

  # -- Nordics AI --

  - name: Lovable
    careers_url: https://jobs.ashbyhq.com/lovable
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/lovable'
    notes: "Stockholm SE. AI app builder (text-to-app), 80+ open roles."
    enabled: true

  - name: Legora
    careers_url: https://jobs.ashbyhq.com/legora
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/legora'
    notes: "Stockholm SE / NYC / London. AI-native legal workspace."
    enabled: true

  - name: Spotify
    careers_url: https://jobs.lever.co/spotify
    notes: "Stockholm SE / NYC / London. Audio platform, large ML/personalization org."
    enabled: true

  - name: Vinted
    careers_url: https://jobs.lever.co/vinted
    notes: "Vilnius LT / Berlin. C2C marketplace unicorn, data science & ML roles."
    enabled: true

  # -- Denmark AI / tech --

  - name: Corti
    careers_url: https://jobs.ashbyhq.com/corti
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/corti'
    notes: "Copenhagen DK. Medical AI (ambient clinical documentation), ML / NLP roles."
    enabled: true

  - name: Pleo
    careers_url: https://jobs.ashbyhq.com/pleo
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/pleo'
    notes: "Copenhagen DK. Spend-management fintech, ML / data / platform roles."
    enabled: true

  - name: Templafy
    careers_url: https://job-boards.greenhouse.io/templafy
    api: https://boards-api.greenhouse.io/v1/boards/templafy/jobs
    notes: "Copenhagen DK. Document-generation / content-enablement platform."
    enabled: true

  # -- Iberia AI --

  - name: Amplemarket
    careers_url: https://job-boards.greenhouse.io/amplemarket
    api: https://boards-api.greenhouse.io/v1/boards/amplemarket/jobs
    notes: "Lisbon PT / Remote. AI-native sales platform."
    enabled: true

  # -- AI / Developer Tools --

  - name: Perplexity
    careers_url: https://jobs.ashbyhq.com/perplexity
    scan_method: websearch
    scan_query: ‘site:jobs.ashbyhq.com/perplexity "AI" OR "Applied AI" OR "Forward Deployed" OR "Engineer"’
    notes: "AI-native search and enterprise AI platform."
    enabled: true

  - name: Clay Labs
    careers_url: https://jobs.ashbyhq.com/claylabs
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/claylabs "AI" OR "Developer Experience" OR "Growth" OR "Engineer"'
    notes: "AI-native GTM and workflow automation platform."
    enabled: true

  - name: Runway
    careers_url: https://job-boards.greenhouse.io/runwayml
    api: https://boards-api.greenhouse.io/v1/boards/runwayml/jobs
    notes: "Generative AI video and creative tooling platform."
    enabled: true

  - name: Hightouch
    careers_url: https://job-boards.greenhouse.io/hightouch
    api: https://boards-api.greenhouse.io/v1/boards/hightouch/jobs
    notes: "Data activation platform with AI agents and solutions engineering roles."
    enabled: true

  - name: WorkOS
    careers_url: https://jobs.ashbyhq.com/workos
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/workos "Solutions Engineer" OR "Developer Success" OR "Engineer"'
    notes: "Developer tools and enterprise-ready auth/platform APIs."
    enabled: true

  # -- Dev Tools / AI Infra (DevRel, SA, FDE focus) --

  - name: Supabase
    careers_url: https://jobs.ashbyhq.com/supabase
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/supabase'
    notes: "Open-source Firebase alternative. Remote-first. Postgres + Auth + Storage. Strong DevRel/DX hiring."
    enabled: true

  - name: Resend
    careers_url: https://jobs.ashbyhq.com/resend
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/resend'
    notes: "Email API for developers. Remote-first. DX/DevRel focused."
    enabled: true

  - name: Clerk
    careers_url: https://jobs.ashbyhq.com/clerk
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/clerk'
    notes: "Auth & user management for developers. SE / TAM / PM roles."
    enabled: true

  - name: Inngest
    careers_url: https://jobs.ashbyhq.com/inngest
    scan_method: websearch
    scan_query: 'site:jobs.ashbyhq.com/inngest'
    notes: "Durable workflow engine for serverless. DX/DevRel/Engineering hiring."
    enabled: true

  - name: PlanetScale
    careers_url: https://job-boards.greenhouse.io/planetscale
    api: https://boards-api.greenhouse.io/v1/boards/planetscale/jobs
    notes: "Serverless MySQL platform. Customer Engineering, Engineering, Sales."
    enabled: true

  - name: Zep AI
    careers_url: https://www.getzep.com/careers
    scan_method: websearch
    scan_query: '"Zep" "Forward Deployed" OR "Developer Relations" OR "Founding" site:workatastartup.com OR site:getzep.com'
    notes: "Context engineering platform for AI agents. YC W24. Lead FDE + DevRel Lead roles. US-only currently."
    enabled: true

  - name: Maxim AI
    careers_url: https://www.getmaxim.ai/careers
    scan_method: websearch
    scan_query: '"Maxim AI" OR "getmaxim" careers "engineer" OR "product"'
    notes: "AI evaluation and observability platform. Bangalore-based, mostly on-site."
    enabled: true

  # ============================================================
  # -- Türkiye / Turkey market
  # ============================================================
  # Tech companies for candidates targeting the Turkish market or
  # Turkish-language postings. Job-portal search queries (Kariyer.net,
  # Yenibiris.com, Secretcv, İşin Olsun, LinkedIn-TR) live in the
  # `search_queries:` section above — that's where the scanner picks
  # them up.
  #
  # NOTE on scanning: Most Turkish companies below do NOT expose a
  # public ATS API (Greenhouse / Lever / Ashby) on their careers_url,
  # so `scan.mjs` will filter them out automatically (it requires a
  # detectable ATS). They are listed here as discovery aids — paste
  # the URL of any specific role into `data/pipeline.md` and run
  # `/career-ops pipeline`, which uses Playwright/WebFetch instead.
  # Only Getir (jobs.lever.co/getir) is currently scannable.
  #
  # Key Turkish market notes:
  #   - Salaries often quoted as NET (after tax/SGK deductions)
  #   - Common benefits: yemek kartı (meal card), özel sağlık sigortası
  #     (private health insurance), servis (company shuttle)
  #   - Probation (deneme süresi): max 2 months by law
  #   - Severance (kıdem tazminatı): 1 month per year of service
  #   - Use modes/tr/ for Turkish-language evaluations and applications
  # ============================================================

  - name: Trendyol
    careers_url: https://www.trendyol.com/kariyerimiz
    scan_method: websearch
    scan_query: 'site:trendyol.com/kariyerimiz OR "trendyol" "Software Engineer" OR "Backend" OR "Frontend" careers'
    notes: "Istanbul (Maslak tech HQ). Turkey's largest e-commerce platform (Alibaba-owned). Large tech org, competitive salaries."
    enabled: false

  - name: Hepsiburada
    careers_url: https://www.hepsiburada.com/kariyer
    scan_method: websearch
    scan_query: '"hepsiburada" careers "Software Engineer" OR "Backend" OR "Frontend" OR "Yazılım"'
    notes: "Istanbul. Major Turkish e-commerce platform. Listed on NASDAQ."
    enabled: false

  - name: Getir
    careers_url: https://jobs.lever.co/getir
    scan_method: websearch
    scan_query: 'site:jobs.lever.co/getir OR "getir" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Ultra-fast delivery unicorn. Exited international markets in April 2024; now Turkey-only."
    enabled: false

  - name: Insider
    careers_url: https://useinsider.com/careers
    scan_method: websearch
    scan_query: '"useinsider" OR "insider" careers "Software Engineer" OR "Backend" OR "Frontend" site:jobs.lever.co OR site:jobs.ashbyhq.com'
    notes: "Istanbul. B2B SaaS marketing platform. Unicorn. Strong remote-friendly culture."
    enabled: false

  - name: Dream Games
    careers_url: https://www.dreamgames.com/careers
    scan_method: websearch
    scan_query: '"dreamgames" OR "dream games" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Mobile gaming company (Royal Match). Strong engineering culture, competitive pay."
    enabled: false

  - name: Peak Games (Zynga)
    careers_url: https://www.peakgames.net/en/career
    scan_method: websearch
    scan_query: '"peak games" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Mobile gaming studio acquired by Zynga. Strong tech team."
    enabled: false

  - name: Yemeksepeti (Delivery Hero)
    careers_url: https://www.yemeksepeti.com/kariyer
    scan_method: websearch
    scan_query: '"yemeksepeti" careers "Software Engineer" OR "Backend" OR "Frontend" OR "Yazılım"'
    notes: "Istanbul. Food delivery platform (Delivery Hero subsidiary). Large tech team."
    enabled: false

  - name: Sahibinden.com
    careers_url: https://www.sahibinden.com/kariyer
    scan_method: websearch
    scan_query: '"sahibinden" careers "Software Engineer" OR "Backend" OR "Frontend" OR "Yazılım Geliştirici"'
    notes: "Istanbul. Turkey's largest classifieds platform. Established tech org."
    enabled: false

  - name: Garanti BBVA Technology
    careers_url: https://www.garantibbvateknoloji.com.tr/kariyer
    scan_method: websearch
    scan_query: '"garanti bbva teknoloji" OR "garantibbvateknoloji" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Fintech/banking tech arm of Garanti BBVA. Large engineering org, competitive benefits."
    enabled: false

  - name: Akbank Tech
    careers_url: https://www.akbank.com/tr-tr/sayfalar/kariyer.aspx
    scan_method: websearch
    scan_query: '"akbank" careers "Software Engineer" OR "Yazılım Mühendisi" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Turkish bank with large in-house tech team."
    enabled: false

  - name: Turkcell
    careers_url: https://kariyer.turkcell.com.tr
    scan_method: websearch
    scan_query: 'site:kariyer.turkcell.com.tr OR "turkcell" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Turkey's largest telecom. Large digital and tech transformation team."
    enabled: false

  - name: Iyzico (PayU)
    careers_url: https://iyzico.com/kariyer
    scan_method: websearch
    scan_query: '"iyzico" careers "Software Engineer" OR "Backend" OR "Frontend" OR "Yazılım"'
    notes: "Istanbul. Fintech / payment infrastructure (PayU subsidiary). Good engineering culture."
    enabled: false

  - name: Papara
    careers_url: https://papara.com/kariyer
    scan_method: websearch
    scan_query: '"papara" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Turkish fintech / digital wallet. Fast-growing, unicorn valuation."
    enabled: false

  - name: Craftgate
    careers_url: https://craftgate.io/career
    scan_method: websearch
    scan_query: '"craftgate" careers "Software Engineer" OR "Backend" OR "Frontend"'
    notes: "Istanbul. Payment orchestration platform. Strong engineering-first culture."
    enabled: false

  - name: n11.com (Doğuş)
    careers_url: https://www.n11.com/kurumsal/kariyer
    scan_method: websearch
    scan_query: '"n11" careers "Software Engineer" OR "Backend" OR "Frontend" OR "Yazılım"'
    notes: "Istanbul. E-commerce marketplace (Doğuş Group)."
    enabled: false

# ── Job Boards ────────────────────────────────────────────────────────
# Multi-employer aggregators. Unlike tracked_companies (one entry = one employer),
# job boards return offers from many companies in a single API call.
# Same title/location filtering and dedup apply.

job_boards:

  # ── SolidJobs (Polish market) ──────────────────────────────────────
  # Public API — no auth needed. Division in URL path, campaign=career-ops for tracking.
  # Available divisions: it, engineering, marketing, sales, hr, logistics, finances, other

  - name: SolidJobs IT
    careers_url: https://solid.jobs/public-api/offers/it?campaign=career-ops
    provider: solidjobs
    enabled: true
    notes: "Polish IT job board — IT division"

  - name: SolidJobs Engineering
    careers_url: https://solid.jobs/public-api/offers/engineering?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Engineering & Production division"

  - name: SolidJobs Marketing
    careers_url: https://solid.jobs/public-api/offers/marketing?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Marketing division"

  - name: SolidJobs Sales
    careers_url: https://solid.jobs/public-api/offers/sales?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Sales division"

  - name: SolidJobs HR
    careers_url: https://solid.jobs/public-api/offers/hr?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — HR division"

  - name: SolidJobs Logistics
    careers_url: https://solid.jobs/public-api/offers/logistics?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Logistics division"

  - name: SolidJobs Finances
    careers_url: https://solid.jobs/public-api/offers/finances?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Finances division"

  - name: SolidJobs Other
    careers_url: https://solid.jobs/public-api/offers/other?campaign=career-ops
    provider: solidjobs
    enabled: false
    notes: "Polish job board — Other specialists division"

  # ── JustJoin.it / NoFluffJobs (Polish tech market) ────────────────────────
  # Public job-board APIs. Disabled by default; enable and rely on the global
  # title_filter/location_filter/salary_filter above for your target profile.

  - name: JustJoin.it
    careers_url: https://justjoin.it/job-offers/all-locations
    provider: justjoin
    enabled: false
    notes: "Polish tech job board — newest public offers"

  - name: NoFluffJobs
    careers_url: https://nofluffjobs.com/pl
    provider: nofluffjobs
    enabled: false
    notes: "Polish tech job board — newest public postings"

  # -- Indonesia job boards (provider: jobstreet / glints) --

  - name: Jobstreet Indonesia
    provider: jobstreet
    api: https://id.jobstreet.com/api/jobsearch/v5/search
    siteKey: ID-Main
    searchKeywords: "data scientist"
    searchLocation: "Jakarta"
    pageSize: 30
    maxPages: 3
    enabled: false
    notes: "Indonesia's largest job board (SEEK group). Uses SEEK v5 JobSearch API — zero tokens."

  - name: Glints Indonesia
    provider: glints
    api: https://glints.com/api/v2-alc/graphql       # searchJobsV3 endpoint
    searchKeywords: "data scientist"
    countryCode: ID
    pageSize: 30
    maxPages: 3
    enabled: false
    notes: "SEA job platform covering ID/SG/MY/VN. Uses searchJobsV3 GraphQL endpoint (v2-alc). WAF blocks non-browser requests — works best from Playwright context."

  # ── HigherEdJobs (university / higher-ed RSS) ────────────────────────
  # Public no-auth RSS category feed. cat_id selects the HigherEdJobs
  # category. Common ones:
  #   68 = Higher Education, 64 = Educational Administration, 22 = Career Development
  # Full list: https://www.higheredjobs.com/rss/

  # - name: HigherEdJobs — Higher Education
  #   provider: higheredjobs
  #   cat_id: 68
  #   enabled: true