Building multi-tenant AI systems with a focus on RAG architecture and agent workflows.
Passionate about performance optimization and open-source contributions.

focus = {
    "current": "Serverless AI Agent · MCP · RAG Architecture",
    "exploring": "LLM optimization, Multi-modal AI",
    "stack": ["Python", "Claude Agent SDK", "AWS Serverless", "MCP"]
}

AI/ML   Python · LangChain · LightRAG · OpenAI · Claude Agent SDK · Vector Databases
Cloud   AWS Lambda · DynamoDB · Bedrock · SAM · Amplify
Tools   Git · Docker · MCP · Cursor

Multi-tenant RAG API powered by LightRAG/RAG-Anything — auto-selects best parser (DeepSeek-OCR/MinerU/Docling) via complexity scoring.
⭐ 47 stars · Recognized by LightRAG author

Serverless AI Agent built on Claude Agent SDK with stateful session persistence via S3+DynamoDB.
🤖 Telegram bot with MCP integration and SQS FIFO async architecture

Lightweight Agentic PDF Search via MCP — Vectorless, reasoning-based document retrieval.
🔍 Inspired by VectifyAI/PageIndex · Native MCP Sampling support

Python SDK for DeepSeek OCR API with intelligent fallback and batch processing.
📦 Published on PyPI

As a member of VectifyAI contributing to Pageindex