AI RAG Vector Search Document Processing

RAG Agent - Document Q&A with Vector Search

AI Agent (Python + LangChain)

Screenshots coming soon

About

Retrieval-Augmented Generation system with dual implementation: lightweight Chroma vector store and enterprise PostgreSQL + PGVector. Answers questions over PDF documents with semantic search.

My Role

Built complete RAG pipeline from PDF ingestion to semantic search to LLM response generation with both Chroma and PostgreSQL backends.

Core Capabilities

PDF loading and semantic chunking (RecursiveCharacterTextSplitter)
Dual vector store: Chroma (lightweight) and PostgreSQL + PGVector (enterprise)
Semantic search retrieval (k=5) with OpenAI embeddings
Chat history persistence via SQLAlchemy
Containerized with Docker Compose

Architecture

LangChain retrieval chain with context-aware prompts
OpenAI text-embedding-3-small for vector indexing
Chroma SQLite backend + PostgreSQL PGVector alternative
Docker Compose for database orchestration

Tech Stack

Python LangChain OpenAI Chroma PostgreSQL PGVector PyPDF Docker Compose SQLAlchemy

Transaction Agent - Email-to-Sheets Pipeline

Stock Agent - PSX Market Data Analysis