Open-source work: RAG systems, multi-agent chatbots, MLOps pipelines, and data engineering tools.
Terminal control panel for MCP servers, published on PyPI as tappy-mcp. Discovers servers across Claude Desktop, Claude Code, and Cursor, edits configs safely with atomic backups, inspects tools over the real MCP protocol, monitors health live, and catches tool rug-pulls with fingerprint pinning.
Real-time AI voice agent for telecom customer support, enabling natural, human-like conversations. Combines LiveKit for real-time streaming, Twilio for telephony, and Deepgram for low-latency speech recognition in an end-to-end ASR, LLM-reasoning, and TTS pipeline.
Production-ready RAG chatbot microservice built with FastAPI, with real-time streaming, LLM caching, advanced vector search, and a seamless knowledge ingestion pipeline, all built for enterprise scale.
AI-powered legal research assistant with a Next.js frontend and LangGraph + FastAPI backend. Uses LLMs to parse and summarize complex legal documents, streamlining research for legal professionals.
Full-stack AI platform that transforms static documents into an interactive knowledge base, with high-performance RAG, persistent document storage, and a modern interface built for enterprise scalability.
End-to-end MLOps solution integrating FastAPI for serving, Docker for containerization, MLflow for experiment tracking, and GitHub Actions for CI/CD, all deployed on AWS infrastructure.
Recommendation engine that merges LLMs with RAG using Gemma-2b to deliver highly personalized content suggestions based on user preferences and retrieval context.
Scalable cloud data engineering pipeline on Azure. Captures sales data, processes it with Azure Data Factory and Databricks, and loads it into a modern cloud data warehouse for analytics.