Summary
Soumyadeep Dey is a Principal Applied Scientist with 11 years of experience designing AI systems that balance accuracy with real-world constraints like cost, latency, reliability, and scale. He specializes in choosing between compact ML models, LLMs, and hybrid architectures to fit the right solution to each problem, and builds evaluation pipelines that align automated LLM-based judgment with human assessment to gate production releases. His PhD from IIT Kharagpur in document image segmentation underpins a track record of shipping production-ready document understanding and vision models using techniques such as quantization-aware training and efficient architecture design. At Microsoft he has led technical direction for dozens of capabilities, creating measurement frameworks that deliver repeatable, production-grade quality signals. He maintains an active research practice with 15+ peer-reviewed publications, and frequently collaborates with academic partners on constrained, applied ML problems.
10 years of coding experience
7 years of employment as a software developer
Master of Engineering (MEng) Computer Science, Master of Engineering (MEng) Computer Science at IIEST, Shibpur
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Indian Institute of Technology, Kharagpur