Summary
S Azam is a research scientist based in Cambridge, MA with nine years of experience at the intersection of representation learning, density estimation, and optimization, now focusing on scalable, privacy-preserving semi/self-supervised methods. Currently at Apple, he develops semi-supervised and federated learning solutions for large-scale end-to-end ASR under real-world constraints like heterogeneity, differential privacy, and distribution shift. His background spans applied ML roles at Zillow and Practo and research at Purdue, where he worked on federated multimodal representations, communication-efficient optimization, and privacy-aware domain alignment. He is particularly skilled at designing algorithms that are communication- and memory-efficient and robust to adversarial perturbations, and has begun integrating deep reinforcement learning with unsupervised learning. Beyond papers and products, he writes about ML practice and research on his personal site and blog, signaling a habit of distilling complex ideas for broader audiences.
9 years of coding experience
7 years of employment as a software developer
Kendriya Vidyalaya Sangathan
Doctor of Philosophy - PhD, Electrical and Computer Engineering, Doctor of Philosophy - PhD, Electrical and Computer Engineering at Purdue University
Bachelor of Technology (B.Tech), Electrical and Electronics Engineering, 8.3/10, Bachelor of Technology (B.Tech), Electrical and Electronics Engineering, 8.3/10 at National Institute of Technology Karnataka
English, Hindi