About

Wei Zhang is a highly skilled Software Engineer with over 20 years of relevant experience. He is currently working as a Software Engineer at Facebook, where he leads the ML Model Platform team. Wei has been instrumental in defining and expanding the team's core vision and evolution strategy. He has built systems and services to tackle challenges in training and deploying some of the world's largest deep learning recommendation models, which have trillions of parameters. Wei has studied scalability limitations of Facebook's ML system architecture and designed and built systems to remove bottlenecks. He has developed a checkpointing system for distributed training reliability, supported fault tolerance to all of Facebook's ML training, and reduced big model training failure rate by 5x. Wei has also delivered transfer learning for ML, which has shortened model iteration lifecycle by 1/4, and enabled cross-domain transfer learning, leading to a 1+% company revenue increase. He has built a distributed model publishing system that has boosted model freshness by 40% for online ML training. Wei has also presented a consolidated, user-friendly model management platform for all Facebook AI/ML models, which manages the full lifecycle of model data at an EB scale. Before joining Facebook, Wei worked as an Engineering Manager at Pure Storage, where he led a team developing space measurement and garbage detection for enterprise storage systems. He also served as a Senior Software Engineer & Tech Lead, Core Infrastructure, where he developed log-structured merge-tree based key-value store for file system metadata and efficient index for very large metadata tables, among other things. Wei has a Doctor of Philosophy (Ph.D.) in Computer Science from the University of California and a Master of Science (M.S.) in Computer Science from Tsinghua University. He also holds a Bachelor of Science (B.S.) in Electrical Engineering and Automation from Wuhan University. Wei's technical skills include Software Engineering, ML, AI, Deep Learning, Infra, QA, Product Manager, DevOps, Research Scientist, and Search. Wei loves to understand and solve hard challenges.