Summary
Yadong Lu is a researcher at Microsoft with nine years of experience focused on efficient pre- and post-training for large language and vision models. He specializes in fine-tuning and customization of diffusion models and LLMs, bringing practical deployment know-how from work on neural network and variable-bit-rate image compression at Qualcomm. With a Ph.D. in Statistics (UC Irvine) and applied mathematics training from NUS and Sichuan University, he combines rigorous quantitative foundations with hands-on systems research. Unusually, he blends compression techniques and model adaptation to reduce compute and data needs for real-world model customization.
10 years of coding experience