Lead AI Engineer at Mercor (Feb 2024 - Present)
Led end-to-end delivery of production AI applications, including structured evaluation of LLM behavior and performance. Improved LLM reasoning accuracy by 18% using structured evaluation frameworks and iterative prompt engineering. Built AI backends that support reliable, observable inference pipelines for high-throughput workloads. • Designed and implemented MCP (Model Context Protocol) servers to enable LLM-powered workflows • Integrated LLM workflows with business data systems (e.g., Bloomberg, QuickBooks, Eightfold, Power BI) • Established engineering standards and mentored team members on AI tooling and deployment • Focused on reliability, observability, and cost efficiency for AI inference