Summary
Haichuan Yang is a staff software engineer and PhD researcher with a decade of experience specializing in efficient deep learning, model compression, sparse optimization, and reinforcement learning. He progressed from publishing foundational work on structured sparsity and energy-constrained compression at top venues (CVPR, ICLR, ICML) to shipping on-device and recommender-system optimizations as a research scientist at Meta and now at Google DeepMind. His research produced practical wins—e.g., automatic joint pruning-and-quantization that compressed AlexNet by 205x without accuracy loss and energy-aware compression reducing inference energy fourfold with minimal accuracy drop. Equally comfortable deriving theory (linear convergence proofs, LP projections for exclusive sparsity) and building production-ready ML systems, he combines rigorous academic training with large-scale industry impact. Based in Beijing with a PhD background from University of Rochester, he quietly bridges cutting-edge research and deployable efficient-AI solutions.
10 years of coding experience
9 years of employment as a software developer
Master's degree Computer Engineering, Master's degree Computer Engineering at Beihang University
Bachelor's degree Software Engineering, Bachelor's degree Software Engineering at Sun Yat-sen University
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at University of Rochester