reranking

Codebase Context gives AI agents understanding of your codebase through semantic code search, team conventions, patterns, and memory, so they use fewer tokens, spend less time, and produce better, more familiar output.

rerank library for easy reranking of results

A modern full-stack personal website and content management system built with Next.js and Hono, featuring a monorepo architecture powered by Turborepo.

RAG System that implement simple and fast. like ragi to make indonesian bread

Enterprise-grade RAG platform — upload PDFs, TXT files, or paste text → asynchronous vector indexing with Qdrant hybrid search → JWT-authenticated chat with streamed SSE answers and source citations. Deployed to Azure Container Apps via GitHub Actions CI/CD.

LLM Banking Compliance Assistant is a production-grade RAG application for querying financial and regulatory documents with high accuracy and verifiable citations. It combines semantic search, cross-encoder reranking, and GPT-4o-mini generation to deliver domain-specific answers to complex banking compliance questions

Product search and recommendations that run on the shopper's device: a 720-product catalog ships as a 1.5 MB signed, content-addressed bundle, then MiniLM semantic search, rank fusion and session-aware reranking all run in the tab. Zero backend calls after sync. 986 tests.

Resilience Hub helps you build consistency: create challenges, log daily notes, and write reflections.

Adaptive RAG engine with sentence-level citation traces, hybrid retrieval, cross-encoder reranking, and evidence memory.

Simple RAG system with a chatbot interface

A full-stack RAG assistant for researching YouTube videos with grounded answers and timestamped evidence.

Vidya — Hybrid memory plugin for OpenClaw agents. LanceDB vector store + Voyage AI embedding/rerank + BM25 keyword search. Multi-scope isolation, auto-capture with LLM judgment, and 400ms avg retrieval.

Hybrid BM25 + vector search CLI for personal knowledge bases. Cross-encoder reranking, SQLite + sqlite-vec, Cloudflare Workers AI.

Workspace-based RAG knowledge search built entirely on Cloudflare Workers — hybrid (vector + BM25) retrieval, cross-encoder re-ranking, multilingual embeddings, and cited streaming chat.

Measured results on retrieval, prompting and agent design — six findings, three contradicting published defaults. Reproducible from a clean clone, every comparison reported with a confidence interval.

A research notebook that answers only from the sources you give it. Add a PDF, a YouTube link or a web page, then ask: every sentence cites the exact page, timestamp or character range behind it. Hybrid retrieval, reranking and a corrective loop that refuses rather than guesses.

Run local reranking models directly in your JavaScript/TypeScript application.

A full-stack Next.js RAG assistant for Udemy course subtitles. Features advanced query translation, hybrid vector/SQL retrieval, Reciprocal Rank Fusion (RRF), LangGraph-based Corrective RAG (CRAG), and input/output guardrails.

Production-ready RAG system for querying academic papers. Upload PDFs, ask questions, get grounded answers with source citations.