Durgesh Rao — AI Researcher

I work on reasoning, reliability, and evaluation of foundation models, with a focus on understanding when and why large language and multimodal models fail. I am a Kaggle Grandmaster and 2× KaggleX Advisor, ranked among the Top 100 global data scientists, which has shaped my approach toward rigorous experimentation, reproducibility, and evaluation beyond leaderboard metrics. My research interests lie at the intersection of LLM reasoning, faithfulness, disagreement-based evaluation, and human-centered diagnostics, where I design causal probes, benchmarks, and evaluation frameworks rather than task-specific models. I have conducted research across academic labs, national research organizations, and industry, including IIIT Bhopal, SAC–ISRO Ahmedabad, DRDO, and currently work as an AI Engineer while pursuing an M.Tech in AI & Data Science. My work emphasizes: 1) Framing research questions, not just optimizing metrics 2) Rigorous evaluation and failure analysis 3) Open, reproducible research artifacts that outlive individual papers My Research interests: LLM reasoning · model faithfulness · disagreement & uncertainty · causal evaluation · multimodal AI · ML systems · Healthcare AI

Stackforce AI infers this person is a Data Scientist specializing in AI and Machine Learning for various industries.

Location: Bengaluru, Karnataka, India

Experience: 1 yr 7 mos