Summary
Sourish Kundu is an Inference Engineer with eight years of hands-on experience building and optimizing machine learning systems that bridge technology and product impact. He has driven significant training and inference speedups at scale—implementing distributed training, mixed precision, and checkpoint strategies at ByteDance—and now focuses on delivering production-ready ML inference at Luma AI. His background spans roles from CTO of a startup to internships at AWS and Capital One, giving him a blend of systems, cloud, and ML engineering expertise. Sourish also co-founded a pandemic-era startup and creates public-facing work on sourish.dev and a technical YouTube channel, signaling a strong bias for shipping and teaching. Notably, he pairs deep performance optimization skills (CUDA/Torch.FX) with product-minded solutions that improve safety and iteration velocity.
8 years of coding experience
5 years of employment as a software developer
Computer Science and Economics Computer Science, Computer Science and Economics Computer Science at University of Wisconsin-Madison
High School Diploma, High School Diploma at Eagan Senior High School
English, Bengali, Hindi