Synapse
Structural context engine for AI agents built with Python, SQLite, MCP, and Tree-sitter. Architected a Git-aware daemon that parses codebases into AST symbols and dependency graphs, built an MCP server for context-bounded code summaries and semantically ranked snippets, engineered asynchronous documentation workers, and designed hybrid vector and SQLite FTS5 retrieval.
RAG Doc Analyzer
Document Q&A AI platform built with Next.js, Express, PostgreSQL, pgvector, and Gemini. Designed a grounded RAG pipeline with Gemini embeddings, pgvector similarity search, PostgreSQL full-text fallback, SSE streaming, Firebase authentication, persistent workspaces, cross-document context orchestration, and asynchronous document ingestion.
LeetSync
Chrome extension and synchronization platform built with TypeScript, Vite, and Manifest V3. Captures LeetCode editor content and metadata, converts submissions into Markdown notes, uses Shadow DOM isolation, and synchronizes solutions to GitHub through the Contents API.