At Turing Enterprises Inc., I build agent benchmarks and evaluation systems for frontier AI labs. I created nearly 30 benchmark tasks and raised an LLM verifier to 93% accuracy and 98% precision against human labels.
At EdgeVerve Systems (an Infosys company), I built multimodal RAG search for contract analysis, raising search accuracy by 70% and cutting retrieval time by 80%. I also build open-source agent and RAG projects, including site2mcp and CORAG.

