An experimental LLM-powered AI pipeline built to explore reliable, structured, and production-oriented AI systems using LangChain, OpenAI, and Ollama, featuring Retrieval-Augmented Generation (RAG) over PDFs, a FastAPI backend, and a Streamlit frontend.
This project focuses on control, validation, and extensibility in AI workflows — moving beyond free-form chatbots toward AI agents that can safely integrate into real-world systems. This demonstrates how to combine cloud LLMs and local LLMs with vector databases to provide intelligent document-aware responses.
- 🔹 OpenAI-powered essay generation
- 🔹 Local LLM support using Ollama (Gemma)
- 🔹 Local embeddings using Ollama embeddings
- 🔹 PDF-based RAG (Retrieval-Augmented Generation)
- 🔹 Vector storage with FAISS / Chroma
- 🔹 FastAPI backend with LangServe
- 🔹 Streamlit interactive UI
- 🔹 Clean modular project structure
- Python 3.10 / 3.11
- LangChain
- Ollama (Gemma, nomic-embed-text)
- OpenAI API
- FAISS / ChromaDB
- Environment-based configuration
- FastAPI
- Streamlit
- PyPDF