Pathway Hackathon: Adaptive RAG Template

This repository provides a ready-to-use Adaptive Retrieval-Augmented Generation (RAG) template using Pathway. It enables you to build, configure, and run a document-based question-answering system with support for both Gemini and OpenAI models.

---

Table of Contents

• Features
• Installation
  • Linux
  • Windows
• Environment Variables (.env)
• YAML Configuration
• Running the Application
• Making Requests (cURL Example)
• References

---

Features

• Document Ingestion: Reads and indexes documents from the data/ folder.
• Flexible LLM Support: Works with Gemini and OpenAI models.
• Configurable via YAML: Easily adjust data sources, models, embedders, and more.
• REST API: Exposes endpoints for question answering.
• Caching: Supports disk and memory caching for efficiency.

---

Installation

For detailed step-by-step instructions, refer to the "Day 1 GenAI Hackathon IIT Jodhpur.pdf.pdf" in this repository.

Linux

# 1. Clone the repository
git clone https://github.com/AISocietyIITJ/pathway_hackathon.git
cd pathway_hackathon

# 2. Create a virtual environment using uv (recommended for speed and reproducibility)
uv venv venv --python 3.10
source venv/bin/activate

# 3. Install all dependencies (including Pathway and extras)
uv pip install "pathway[all]"

Windows

# 1. Clone the repository
git clone https://github.com/AISocietyIITJ/pathway_hackathon.git
cd pathway_hackathon

# 2. Build the Docker image
docker build -t my-pathway-app .

# 3. Run the container (maps your current directory to /app in the container)
docker run -it --rm -v %cd%:/app -p 8008:8000 my-pathway-app

---

Environment Variables (.env)

Create a .env file in the root directory with the following structure:

# For Gemini (Google) API
GEMINI_API_KEY="your-gemini-api-key"

# For OpenAI API (if using OpenAI models)
OPENAI_API_KEY="your-openai-api-key"

• Get your Gemini API key from Google Cloud Console.
• Get your OpenAI API key from OpenAI.

---

YAML Configuration

The application is configured using YAML files (app.yaml or app_hydrid_retriever.yaml). These files define:

• Data Sources: Where your documents are loaded from (e.g., local data/ folder).
• LLM Model: Which language model to use (Gemini, OpenAI, etc.).
• Embedder: For text embeddings (GeminiEmbedder, OpenAIEmbedder).
• Splitter & Parser: How documents are chunked and parsed.
• Retriever: How documents are indexed and retrieved (BruteForce, Hybrid, etc.).
• Server Settings: Host, port, caching, and error handling.

To modify the pipeline, edit the relevant YAML file.
For more details, see comments in app.yaml and Pathway YAML documentation.

---

Running the Application

Make sure your .env and YAML files are set up.

# Run the main application
python main.py

• By default, the server will start at http://0.0.0.0:8000.
• You can change the host and port in the YAML file.

---

Making Requests (cURL Example)

To query the API, use the following cURL command:

curl --location 'http://localhost:8000/v2/answer' \
  --header 'accept: */*' \
  --header 'Content-Type: application/json' \
  --data '{
    "prompt": "Give me highlights of Q3 Financial Summay"
}'

• Add your query in 'prompt' field
• The server will return a JSON response with the answer.

---

References

• Pathway Documentation
• YAML Configuration Guide
• Day 1 GenAI Hackathon IIT Jodhpur.pdf.pdf (for detailed setup)

---

Happy Hacking!

---