Model and pipeline optimization for large generative models

Deep learning model and pipeline optimization. Acquired by NVIDIA.

OmniML empowers users to design, optimize, shrink, and deploy

advanced machine learning models to resource-constrained

hardware devices, focusing on mobile, automotive, robotics, and IoT applications.

OmniML’s AI software technology enables enterprises to produce small, fast, and energy-efficient machine learning models for on- device AI.

OmniML solves the mismatch between AI applications and hardware deployments to make AI more accessible everywhere.

OmniML was founded in 2021 by Dr. Song Han, MIT EECS Professor and experienced start-up founder, Dr. Di Wu, former Facebook engineer, and Dr. Huizi Mao, co-inventor of the “deep compression” technology coming out of Stanford.

Commerce (Marketplace, Shops and Jobs) Machine Learning Infra Systems.

Worked on autonmous driving perception algorithms, and on-vehcicle/backend infra systems.

Responsible for technical innovation and development of FlashBlade. One of the major contributors to the flash management subsystem, the HA subsystem, the metadata subsystem and the snapshots subsystem.

FlashBlade is an all-flash scale-out distributed file and object storage product designed from the ground up to meet the demands of modern applications.