Summary
Varad Bhatnagar is an Applied Scientist specializing in aligning large language models with human preferences, currently improving Microsoft Enterprise Copilot within the Turing team after pivotal work on AI-for-code at IBM Research. Over nine years he has bridged research and productization—building training and evaluation pipelines for models up to 20B parameters, contributing to WatsonX Code Assistant, and filing multiple patents. His academic background (M.Tech, IIT Bombay) includes automated fact-checking and multimodal NLP, and he has published in venues like COLING and INTERSPEECH. Pragmatic yet research-driven, he combines deep systems engineering (distributed training, dynamic data mixing) with a focus on usable interactions such as generating engaging follow-up questions and tooling for reasoning and tool use.
9 years of coding experience
2 years of employment as a software developer
Indian Institute of Technology Bombay
Bachelor of Engineering - BE, Information Technology, 9.17/10, Bachelor of Engineering - BE, Information Technology, 9.17/10 at Army Institute of Technology, Pune