I've built production LLM systems and AI data pipelines at RapidNode, where I own services from prototype through deployment and on-call operation.
I built a Python and LangChain AI agent with PGVector retrieval that automated airdrop discovery from more than 10,000 daily tweets, replacing manual analyst curation. I also architected RapidNode's enterprise AI data pipeline, a revenue-driving product built with NestJS and Python microservices.
My backend work includes cost-optimized LLM scoring, RabbitMQ-based async processing for 10,000+ daily events, and AWS observability with Prometheus and Grafana. Previously at Clamp, I developed ERC-4337 smart contract accounts that reduced user gas costs by 60% and helped launch mobile and rewards experiences.
I've also shipped a production DeFi swap and onramp widget and built job-search tools using LLM evaluation, hybrid pgvector search, PostgreSQL, and multi-turn conversational workflows.
