At Blockchain Relations, I built Python and LLM agents that automate prospect research, saving 12 hours of manual work each week. I also built a GTM stack connecting HubSpot, Clay, Stripe, and PostHog, and used experiments and product-event signals to improve conversion and revenue.
At CryptoLiveLeak, Inc., I built multi-agent anomaly detection across 10M+ data points and RAG pipelines powering a chatbot that resolved 80% of routine queries. I improved LLM accuracy while reducing token costs and inference latency, and introduced evaluation, observability, and guardrails for production systems.

