Summary
Ahmed Humayun is a research scientist focused on how data (real and synthetic) and training objectives shape memorization, generalization, and robustness in large-scale image, video, and text foundation models. With eight years of experience spanning academic research at Rice University, industry research roles at Google, and co-founding Bengali.AI, he blends theoretical analysis—particularly function geometry of training dynamics—with practical work on billion-parameter models. He aims to make theory transferable to practice by developing visualization and characterization tools that reveal post-training behavior and improve generative model utility. Based in New York, Ahmed combines deep learning PhD training with hands-on engineering across research and startup environments, bringing an unusual emphasis on interpretable, theoretically inspired frameworks that scale to real-world foundation models.
8 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD Deep Learning, Doctor of Philosophy - PhD Deep Learning at Rice University
Bachelor of Science (B.Sc.) Electrical and Electronics Engineering, Bachelor of Science (B.Sc.) Electrical and Electronics Engineering at Bangladesh University of Engineering and Technology