flashrank

Production-grade Corrective RAG (CRAG) & Hybrid Retrieval engine combining BM25 lexical search, Dense Vectors, Reciprocal Rank Fusion (RRF), and ONNX Cross-Encoder reranking.

Supplementary benchmarks for Making Legacy Knowledge Searchable with RAG

Two-stage hybrid search engine combining PostgreSQL pgvector (HNSW), full-text search (GIN), Reciprocal Rank Fusion (RRF), and local FlashRank cross-encoder reranking.

Agentic RAG with LangGraph, guardrails, FlashRank reranking, and page-accurate citations. Runs locally on Groq + embedded Qdrant.

Production-grade document intelligence and RAG platform. Users upload PDFs, DOCX, PPTX, XLSX, CSV, etc., then ask questions and receive cited answers. Uses hybrid vector + keyword retrieval, reranking, agentic retrieval retries, conversation memory, and LLM evaluation.

Ultra-lightweight (~40MB RAM), Zero-GPU Two-Stage RAG engine powered by SQLite FTS5, local FlashRank cross-encoder, and semantic caching.

A progressive, script-based journey through Retrieval-Augmented Generation, from basic RAG to agentic, self-reflecting systems. Free-tier cloud stack, zero local model downloads.

AI-powered hybrid search engine for arXiv papers. Natural language queries → arXiv API → FAISS + BM25 → RRF fusion → FlashRank reranking → LLM synthesis.

Context-Oriented Institutional Knowledge Management System (IKMS). Features Adaptive Hybrid RRF Retrieval (BM25 + ChromaDB), FlashRank Cross-Encoder re-ranking, Groq Llama-3.3 grounded Q&A with verifiable citations, SHA-256 tamper-evident audit ledger, and OWASP ASVS 4.1 RBAC multi-tenancy.

Agentic RAG application for technical document Q&A, with LangGraph query planning, Qdrant vector search, FlashRank reranking, Gemini embeddings, Portkey/Groq routing, and NeMo input guardrails. Includes local document ingestion, FastAPI and Streamlit interfaces, Logfire/LangSmith tracing, and RAGAS evaluations.

Enterprise RAG pipeline combining Dense Vector Search, BM25 Lexical Retrieval, and Cross-Encoder Re-ranking via Groq LPU.

AI customer-support backend: LangGraph intent router + 5 specialist nodes, hybrid RAG (Gemini embeddings, pgvector/Qdrant, FlashRank reranking), Redis caching, FastAPI REST + WebSockets, Docker.

Enterprise RAG over the 686-page Income Tax Act 2025. Hybrid Search (BM25 + Vectors), Query Rewriting, and Cross-Encoder Reranking.

Local ONNX cross-encoder reranking for JavaScript/TypeScript. Zero API costs, framework-agnostic, with a Vercel-style API.

Enterprise RAG Workflow Application — PDF upload, semantic search, FlashRank reranking, Chain-of-Thought reasoning, and PDF report generation using FastAPI + React + ChromaDB

Production-grade RAG chatbot at $0/month: FastAPI, hand-rolled two-stage retrieval (NumPy cosine → FlashRank cross-encoder), refusal floor, SSE streaming, abuse protection. No LangChain, no vector DB.

A low-latency full-stack RAG chatbot built with FastAPI, React, PostgreSQL, and pgvector, featuring PDF/DOCX ingestion, hybrid retrieval, grounded chat, streaming responses, and optional FlashRank reranking.

Enterprise Document Intelligence & Advanced RAG System

AI-Powered NLP Dataset Generator & Semantic Search Platform | FastAPI, Next.js, Redis Vector DB, FlashRank, Docker