Deployed Live at https://multi-agent-platform-alpha.onrender.com

Or locally after cloning and supplying Groq API key

uvicorn app:app --reload --port 8000

Fully dockerized and continuosly updated

Standalone Streamlit NL2SQL and FileSystem Agent + Semantic Web Searcher with tool binding demos: https://rodneyfinkel-text-analysis-pydantic-streamlit-app4-h4ppca.streamlit.app/

Data Pipeline Matrix

LangGraph supervisor–worker architecture with a ReAct agent as a subgraph node

##SCREENSHOTS

CHAT UI FOR Graph workflows

OR Single node access for dedicated Agent

STAND ALONE AGENTS

EMAIL & DATA ASSEMBLY AGENT (PARQUET attachments for DB query results)

FastAPI Swagger UI for simple API testing

GROQ_API_KEY is needed inside a .env file Clone the repository

Create a virtual environment python -m venv venv

Activate the virtual environment On macOS/Linux: source venv/bin/activate On Windows: .\venv\Scripts\activate

Install all required libraries using the requirements.txt file provided in the repository: pip install -r requirements.txt

Launch the Streamlit server to view the app in your browser: streamlit run streamlit_app4.py

STREAMLIT ReACT Agent for Filesystem and NL2SQL

Why Llama 3.3 70B Versatile?

It is specifically tuned to excel at JSON mode and Function Calling 128K Context Window: Supports a very large 128,000-token context. No need to build chunking or map-reduce logic.

Using Pydantic replaces traditional prompt-engineering for output formatting by providing a schema contract that forces the LLM to return valid, structured data. This was chosen to eliminate unpredictable text chatter and ensure type-safe validation. Moving logic from raw strings to Python objects makes the app becomes more robust.