#reranking

Hands-on Generative AI: 15 RAG modules from document loaders to guardrails, plus Papeer — a research paper assistant with citations and claim verification. Also LangGraph workflows and LangSmith tracing.

AI document assistant with a production RAG pipeline — hybrid BM25 + semantic search, cross-encoder reranking, and streaming answers. Built with FastAPI, Streamlit, ChromaDB, and DeepSeek LLM. Fully Dockerized and deployed on Railway.

Local-first GraphRAG: Milvus hybrid (dense+BM25) + Neo4j local-search + cross-encoder rerank + contextual retrieval. Local-dev runs end-to-end on Ollama with no paid APIs; CI integration tier uses Ollama Cloud + OpenAI embeddings.

Self-hosted knowledge base for AI agents: ingest Markdown into Qdrant with H1-H...-aware chunking, hybrid (dense+sparse) search and cross-encoder reranking.

A practical, first-principles handbook on production RAG & agentic systems — RU/EN/SK, built with Docusaurus.

多模态内容理解与工具编排 Agent：Chinese-CLIP 图文共享向量空间 + BM25/RRF 混合检索 + cross-encoder 重排（真实文本查询 hit@10 0.668→0.918）+ Schema 约束工具调用 + 可溯源生成，含 MCP、多轮会话记忆、ReAct 编排与分层离线评测。 | Multimodal agent: Chinese-CLIP retrieval, BM25+RRF hybrid search, cross-encoder reranking, schema-constrained tool calling, MCP server.

Benchmark LLM-native structured output against TypeSafe Jev across latency, cost, and judgment quality.