chore: import upstream snapshot with attribution

deploy-agentic-rag/.env.example

@@ -0,0 +1,2 @@
+SERPER_API_KEY="your-api-key-here"
+OPENAI_API_KEY="your-api-key-here"

deploy-agentic-rag/README.md

@@ -0,0 +1,43 @@
+# 100% private Agentic RAG API
+
+This is a simple API that uses CrewAI and LitServe to create a 100% private Agentic RAG API.
+
+## How to use
+
+1. Clone the repo
+2. Install the dependencies:
+
+```bash
+pip install crewai crewai-tools litserve
+```
+
+Download Ollama and run the following command to download the Qwen3 model:
+
+```bash
+ollama pull qwen3
+```
+
+3. Run the server:
+
+```bash
+python server.py
+```
+
+4. Run the client:
+
+```bash
+python client.py --query "What is the Qwen3?"
+```
+
+---
+
+## 📬 Stay Updated with Our Newsletter!
+**Get a FREE Data Science eBook** 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. [Subscribe now!](https://join.dailydoseofds.com)
+
+[![Daily Dose of Data Science Newsletter](https://github.com/patchy631/ai-engineering/blob/main/resources/join_ddods.png)](https://join.dailydoseofds.com)
+
+---
+
+## Contribution
+
+Contributions are welcome! Please fork the repository and submit a pull request with your improvements.

deploy-agentic-rag/client.py

@@ -0,0 +1,33 @@
+# client.py
+import requests
+import json
+import argparse
+import time
+# replace this URL with your exposed URL from the API builder. The URL looks like this
+SERVER_URL = 'http://0.0.0.0:8000'
+
+def main():
+    parser = argparse.ArgumentParser(description="Send a query to the LitServe server.")
+    parser.add_argument("--query", type=str, required=True, help="The query text to send to the server.")
+    
+    args = parser.parse_args()
+    
+    payload = {
+        "query": args.query
+    }
+    
+    try:
+        response = requests.post(f"{SERVER_URL}/predict", json=payload)
+        response.raise_for_status()  # Raise an exception for bad status codes
+        
+        result = response.json()['output']['raw']
+        for token in result.split():
+            print(token, end=" ", flush=True)
+            time.sleep(0.05)
+
+        # print(json.dumps(result, indent=2))
+    except requests.exceptions.RequestException as e:
+        print(f"Error sending request: {e}")
+
+if __name__ == "__main__":
+    main()

deploy-agentic-rag/server.py

@@ -0,0 +1,64 @@
+# rename .env.example to .env and add the following:
+# SERPER_API_KEY=your_serper_api_key
+# OPENAI_API_KEY=your_openai_api_key
+
+from crewai import Crew, Agent, Task, LLM
+import litserve as ls
+from crewai_tools import SerperDevTool
+
+# If you'd like, you can use a local LLM as well through Ollama. Do this:
+# ollama pull qwen3 in the command line.
+
+# Uncomment the following line and also the llm=llm line in the Agents definitions.
+# llm = LLM(model="ollama/qwen3")
+
+class AgenticRAGAPI(ls.LitAPI):
+    def setup(self, device):
+        researcher_agent = Agent(
+            role="Researcher",
+            goal="Research about the user's query and generate insights",
+            backstory="You are a helpful assistant that can answer questions about the document.",
+            verbose=True,
+            tools=[SerperDevTool()],
+            # llm=llm
+        )
+
+        writer_agent = Agent(
+            role="Writer",
+            goal="Use the available insights to write a concise and informative response to the user's query",
+            backstory="You are a helpful assistant that can write a report about the user's query",
+            verbose=True,
+            # llm=llm
+        )
+        
+        researcher_task = Task(
+            description="Research about the user's query and generate insights: {query}",
+            expected_output="A concise and informative report about the user's query",
+            agent=researcher_agent,
+        )
+
+        writer_task = Task(
+            description="Use the available insights to write a concise and informative response to the user's query: {query}",
+            expected_output="A concise and informative response to the user's query",
+            agent=writer_agent,
+        )
+        
+        self.crew = Crew(
+            agents=[researcher_agent, writer_agent],
+            tasks=[researcher_task, writer_task],
+            verbose=True,
+        )
+
+    def decode_request(self, request):
+        return request["query"]
+
+    def predict(self, query):
+        return self.crew.kickoff(inputs={"query": query})
+
+    def encode_response(self, output):
+        return {"output": output}
+
+if __name__ == "__main__":
+    api = AgenticRAGAPI()
+    server = ls.LitServer(api)
+    server.run(port=8000)