At AirDawg Labs, I design and evaluate terminal-based benchmark tasks for AI agents, focused on realistic, multi-step engineering and system workflows. I create Terminal-Bench/Harbor tasks with executable tests, solution scripts, and reproducible evaluation workflows.
I analyze automated test results, CI and reviewer feedback, and LLM-as-a-Judge signals to find coverage gaps, ambiguity, and failure modes. I harden tasks using Linux, Docker, Git, and AI coding agents.
As a Backend Developer Intern at Infobell IT Solutions Pvt. Ltd., I worked on B4iGo, testing core modules and tracing edge-case failures. I also developed backend systems through engineering projects, including an e-commerce microservices backend with Go, gRPC, and a GraphQL API Gateway.
