PeopleofSuperintelligence

Researcher

Minjia Zhang

Minjia Zhang researches efficient AI systems, from CPU-based recurrent-network serving to large-scale language-model training and inference. He earned a Ph.D. in computer science and engineering at Ohio State in 2016 and coauthored DeepCPU at Microsoft in 2018. In 2022, he worked on DeepSpeed-MoE for mixture-of-experts training and inference; his 2024 ICLR paper on adaptive KV-cache compression earned an Outstanding Paper Award honorable mention. He joined UIUC as an assistant professor in the Siebel School of Computing and Data Science in 2024 after seven years at Microsoft Research Redmond and WebXT as a principal researcher and technical lead, and he received the Amazon Research Award and Google ML and Systems Junior Faculty Award in 2025. He directs UIUC's Supercomputing System and AI Lab, where his 2026 ICLR work AutoSP extends context lengths for language-model training.

Badges

Timeline

  1. 2016

    An Ohio State PhD

    He completes a Ph.D. in computer science and engineering at Ohio State in 2016.

  2. 2018

    DeepCPU on CPUs

    In 2018, he coauthors DeepCPU, a CPU library for serving recurrent neural networks. The paper reports an order-of-magnitude gain in latency and efficiency over existing deep-learning frameworks.

  3. 2022

    DeepSpeed-MoE at Microsoft

    In 2022, he is an early member of Microsoft's DeepSpeed team and coauthors DeepSpeed-MoE, an end-to-end mixture-of-experts training and inference system. The paper reports a fivefold training-cost saving for language models.

  4. 2024

    An ICLR KV-cache paper

    In 2024, he coauthors Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs. The paper introduces adaptive compression to reduce the memory footprint of LLM inference and earns an ICLR Outstanding Paper Award honorable mention.

  5. 2024

    Joining UIUC faculty

    At UIUC, he joins the Siebel School of Computing and Data Science as an assistant professor in 2024, after seven years at Microsoft Research Redmond and WebXT as a principal researcher and technical lead.

  6. 2025

    Research awards

    In 2025, he receives the Amazon Research Award and the Google ML and Systems Junior Faculty Award.

  7. 2026

    AutoSP for long contexts

    AutoSP is his ICLR 2026 work on sequence-parallel execution for long-context training. It converts standard training code and increases maximum trainable context length by up to 2.7 times on NVIDIA GPUs and 2.5 times on AMD GPUs.

  8. As of October 2026

    Directing SSAIL at Illinois

    As of October 2026, he is an assistant professor in UIUC's Department of Computer Science and director of the Supercomputing System and AI Lab (SSAIL).

Career

  1. Since 2024University of Illinois Urbana-ChampaignAssistant Professor
  2. Until 2024Microsoft ResearchPrincipal Researcher and Technical Lead
  3. 2022–2024MicrosoftEarly member of the DeepSpeed team, Principal Researcher and Technical Lead

Education

  1. 2016Ohio State UniversityPh.D.

From the biography.

  1. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Minjia Zhang. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.
  2. 16% of AI career moves in the 2020s went from a university to a companyIncludes Minjia Zhang. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes University of Illinois Urbana-Champaign. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  4. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorateIncludes University of Illinois Urbana-Champaign. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorate. Across 2,116 people with both an undergraduate degree and a doctorate recorded, we trace which undergraduate schools feed which doctoral programs, how often students stay at the same school, and how paths differ by birth-country region.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Minjia Zhang yet. Send us a free-licensed photo.

  1. OhioLINK ETD: Zhang, Minjia etd.ohiolink.edu. Checked 21 October 2026.
  2. DeepCPU: serving RNN-based deep learning models 10x faster - Microsoft Research microsoft.com. Checked 21 October 2026.
  3. DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scale - Microsoft Research microsoft.com. Checked 21 October 2026.
  4. Minjia Zhang minjiazhang.github.io. Checked 21 October 2026.
  5. Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs - Microsoft Research microsoft.com. Checked 21 October 2026.
  6. ICLR 2024 Awards iclr.cc. Checked 21 October 2026.
  7. Minjia Zhang | Electrical & Computer Engineering | Illinois ece.illinois.edu. Checked 21 October 2026.