At Neuroscale AI, I engineered the arbitrator domain, a core AI pipeline that unified candidate sourcing, resume screening, and role-based ranking into one workflow, reducing manual screening time by up to 80%. I also built a parallel processing pipeline that achieved hundreds of thousands of LLM evaluations in under a minute.
I designed and owned the outreach sequence system, the platform's largest feature, and built a Python and FastAPI service to orchestrate AI agents and background search tasks. To support a candidate dataset growing to billions of rows, I introduced sharding strategies and a dual-database layer with PostgreSQL and ClickHouse.
Across roles at Rightway, Impulso, and other teams, I've worked on Node.js and React products, distributed systems, and domain-driven design. At Rightway, I reduced integration test time from over five seconds to under one second per test; at Impulso, I designed an event-driven billing CRM platform that processed millions of financial transactions per day in real time.
