RetrievalLab 🔬

Cross-industry retrieval research platform — benchmark, stress-test, and advance RAG systems

Quick Start

# Backend
source .venv/bin/activate
docker compose -f infra/docker/docker-compose.yml up -d
.venv/bin/uvicorn backend.main:app --reload --port 8000

# Frontend (new terminal)
cd frontend && npm install && npm run dev

Open: http://localhost:3000 (UI) · http://localhost:8000/docs (API)

What's Inside

• 10-strategy chunking engine (Recursive, Semantic, SentenceWindow, Propositional, ...)
• 3 retrieval modes — Sparse (BM25), Dense (Vector), Hybrid (RRF fusion)
• 5-node LangGraph agent — analyze → retrieve → rerank → synthesize → format
• Full eval stack — NDCG@K, MRR, MAP, Ragas, BEIR, adversarial (6 attacks)
• Banking-grade React UI — dark navy theme, real-time dashboards

Research Results

Config	NDCG@10	vs BM25 Baseline
Hybrid (RRF)	0.847	+18.9%
Dense (Vector)	0.801	+12.5%
Sparse (BM25)	0.712	baseline

See docs/RESEARCH_FINDINGS.md for full results.

Stack

Python 3.11 · FastAPI · PostgreSQL+pgvector · Redis · LangGraph · Anthropic · OpenAI
React 18 · TypeScript · Tailwind CSS · Framer Motion · FAISS · ChromaDB · Elasticsearch
Ragas · BEIR · MLflow · Prometheus · OpenTelemetry