Agentic AI Trainer
Optimized autonomous AI agents for repository-level software engineering tasks across Python, TypeScript, Go, and SQL. Engineered and evaluated 'Gold Trajectories' for agentic workflows, focusing on multi-step reasoning, tool-call accuracy (CLI, Linters, Debuggers), and autonomous bug resolution within open-source codebases. Developed comprehensive alignment rubrics to grade agent planning, ensuring adherence to language-specific best practices and architectural constraints.