LexAI: A Legal Research Assistant for Indian Law

LexAI is a retrieval-augmented generation (RAG) system for answering legal questions about Indian law. Instead of guessing, it finds the relevant passages in 18+ Indian statutes and Supreme Court judgments and builds its answer from them, citing the sources.

What I built:

A document ingestion pipeline with semantic chunking and metadata-aware indexing

Hybrid retrieval that combines FAISS (semantic search) and BM25 (keyword search), merged with Reciprocal Rank Fusion

CrossEncoder reranking and HyDE query rewriting to improve answer relevance

An async FastAPI backend that streams answers live (Server-Sent Events), remembers the conversation within a session, and returns source citations

A React and Tailwind CSS frontend with real-time streaming, Markdown rendering, and interactive citation panels