Description

ComplianceIQ is an enterprise-grade RAG (Retrieval-Augmented Generation) assistant built to help legal and compliance teams query, analyze, and stay current on regulatory documents - without manually combing through hundreds of pages.

The Problem
Legal compliance teams spend hours manually cross-referencing policy documents, regulations, and internal guidelines. Existing tools either lack accuracy on domain-specific legal language or don't scale to enterprise document volumes.

The Solution
ComplianceIQ combines hybrid search, advanced retrieval techniques, and multi-provider LLM routing to deliver accurate, cited answers from large legal document sets - with full audit traceability for compliance-sensitive environments.

Key Features

• Hybrid Search with RRF (Reciprocal Rank Fusion) - combines dense vector search and keyword search for higher retrieval accuracy

• Cross-Encoder Reranking - re-scores retrieved passages for relevance before generation

• HyDE (Hypothetical Document Embeddings) - improves retrieval on complex/ambiguous legal queries

• Multi-Provider LLM Routing - automatic failover and key rotation across LLM providers for reliability

• Google OAuth with Role-Based Access Control - secure, enterprise-ready authentication

• SQLite Audit Trail - full query and document access logging for compliance requirements

• Scheduled Re-Analysis - automated periodic document re-checks via APScheduler

• Report Export - generate PDF/Word compliance reports directly from query sessions

Tech Stack
Built with Python, using Qdrant as the vector database, LangChain for document chunking and orchestration, and Streamlit for the interface. Deployed on Railway (backend) and Streamlit Community Cloud (frontend), with SQLite powering the audit and governance layer.

Status

Actively developed - built from local prototype through full cloud deployment