At Turing, I build reinforcement-learning environments and evaluation harnesses for frontier-lab and enterprise programs. I delivered a recruiting-ecosystem environment through client acceptance and developed offline, containerized environments where computer-use agents’ actions are verified against application state.
I designed an evaluation framework spanning 130 cross-application tasks and roughly 8,700 programmatic assertions, with rubric criteria and an LLM QA-reviewer agent. I also created Playwright and state-diff tests to re-test rejected tasks and identify recurring defect families.
Earlier, at KNS Technologies, I built a RAG assistant over a 10,000+ document knowledge base and ML pipelines for student performance data. My experience also includes enterprise virtual assistants, CRM integrations, and independently shipped projects such as a Telugu-English voice agent and an AML/CTF knowledge assistant.
