Heterogeneous Extreme Computing (HEX) Group

The mission of our group is to “deliver affordable and reliable AI applications to clients”. In this end, our research directions cover the whole AI stack as follows. Our group has a range of superior research achievements published in top-tier conferences, and also successful technique transfers to productions including Office, Bing, Teams, and so on.

• Affordable AI model design (model compression and optimization, quantization, etc.)
• Efficient AI systems/compilers (inference systems for heterogeneous hardware, resource management and scheduling, tensor compilers for new hardware, etc.)
• Big science on cloud and parallel algorithm (Cloud for AI+Science, HPC+AI model & algorithm design etc.)
• Hardware design for AI and AI for hardware design (AI accelerator design, AI accelerator and model co-design, AI for electronic design etc.)
• Learning on the edge (Foundation model fine tuning, distributed learning, continuous learning, collaborative learning etc.)
• New applications and scenarios (AIoT, AR, VR, gaming, 5/6G etc.)

The Principal Research Manager of HEX group is Ting Cao, and welcome to join us as Researchers, Intern students, or Research Engineers.