Deepan Adak

I am an Applied AI Scientist at Microsoft, working on finetuning Small Language Models with RLHF and custom reward policies for data security compliance. I have a strong background in machine learning engineering and research spanning defense, healthcare, and computational chemistry.

Previously, I was a Founding Engineer at Zenteiq.ai (incubated at IISc, Bangalore), where I built high-throughput LLM serving infrastructure, multiagent RAG pipelines, and real-time face recognition systems. I also conducted research as a Project Associate at the AIREX Lab, IISc, collaborating with DRDO on radar intercept classification and defense surveillance systems.

My research has been published and accepted at venues including NeurIPS 2025, and I have work under review at COLM 2026, and leading journals. My interests span Vision-Language Models, molecular property prediction, medical image analysis, and scalable ML systems.