PeopleofSuperintelligence

Lab leader

Zain Asgar

Zain Asgar builds systems for AI inference as co-founder and CEO of Gimlet Labs and an adjunct professor of computer science at Stanford University. He wrote GPU Energy Modeling and Analysis at Stanford in 2015 and worked on apparel recognition features at Google that became part of Google Lens. In 2018, he co-founded Pixie Labs, which built tools for monitoring applications and joined New Relic through an acquisition in 2020. His 2025 research described how to divide AI agent workloads across different hardware and introduced KForge, a system that uses language model agents to write and improve GPU kernels. In September 2026, Gimlet announced a $300 million Series B and a partnership to bring Cerebras processors into its inference cloud.

Now at
Gimlet Labs and Stanford University
Works
18
Citations
118
h-index
6
Badges

Publication numbers from OpenAlex, as of 27 October 2026.

Timeline

  1. June 2015

    GPU energy research

    At Stanford, he completes the thesis GPU Energy Modeling and Analysis in June 2015. He earns a PhD at the university.

  2. 2018

    Co-founding Pixie Labs

    He co-founds Pixie Labs with Ishan Mukherjee in 2018 to build tools for exploring and monitoring applications. His earlier work in Google’s machine intelligence group includes apparel recognition features that become part of Google Lens and machine learning on devices; he also serves as an entrepreneur in residence at Benchmark before founding Pixie.

  3. 2020

    Pixie joins New Relic

    Pixie launches its public beta in October 2020, giving developers access to application performance data. In December 2020, New Relic signs an agreement to acquire Pixie Labs.

  4. 2021

    Open source at New Relic

    After New Relic acquires Pixie Labs in December 2020, he announces Pixie’s release as open source software under the Apache 2.0 license in May 2021. He serves as general manager and group vice president for Pixie and open source at New Relic in 2021. His earlier industry work includes data and AI/ML systems at Google AI, Trifacta and NVIDIA.

  5. 2022

    Debugging hardware generators

    With Keyi Zhang and Mark Horowitz, he coauthors Bringing Source-Level Debugging Frameworks to Hardware Generators in 2022. Their framework connects hardware generators to software debugging tools, allowing designers to set breakpoints and inspect variables in the original code.

  6. July 2025

    Systems for AI agents

    In July 2025, he coauthors Efficient and Scalable Agentic AI with Heterogeneous Systems with Michelle Nguyen and Sachin Katti. They describe a system that divides AI agent workloads among CPUs and accelerators, with cost models, compilation and dynamic scheduling to assign work to different hardware.

  7. November 2025

    KForge generates GPU kernels

    He coauthors KForge: Program Synthesis for Diverse AI Hardware Accelerators in November 2025. KForge uses two collaborating language model agents to generate and improve GPU kernels: one writes programs, and one analyzes performance to guide changes. The work demonstrates the method on NVIDIA CUDA and Apple Metal.

  8. June 2026

    KForge across AI accelerators

    In June 2026, he coauthors KForge: LLM-Driven Cross-Platform Kernel Generation for AI Accelerators. The team evaluates the framework on NVIDIA B200 and Intel Arc B580 hardware, using compilation, correctness tests and performance analysis to improve generated kernels.

  9. September 2026

    Gimlet funding and Cerebras

    Gimlet announces a $300 million Series B led by Andreessen Horowitz in September 2026. In September 2026, he announces a partnership with Cerebras to bring its wafer-scale processors into Gimlet Cloud; the companies plan 100 megawatts of inference capacity.

  10. As of October 2026

    Gimlet CEO and Stanford professor

    As of October 2026, he is co-founder and CEO of Gimlet Labs and an adjunct professor of computer science at Stanford University. Gimlet builds AI inference systems that divide workloads across different types of processors.

Career

  1. NowGimlet LabsCo-founder and CEO
  2. NowStanford UniversityAdjunct professor of computer science
  3. From 2018Pixie LabsCo-founder
  4. BenchmarkEntrepreneur in residence
  5. New RelicGeneral manager and group vice president for Pixie and open source

Education

  1. Stanford UniversityPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Checked career sources: chipgen.stanford.edu, mlsys.stanford.edu, blog.px.dev.

Colleagues

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

Sources and credits

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

  1. Alumni theses chipgen.stanford.edu. Checked 27 October 2026.
  2. Stanford MLSys Seminar – Stanford MLSys Seminar Series mlsys.stanford.edu. Checked 27 October 2026.
  3. Announcing Pixie’s Public Beta Launch and Series A | Pixie Labs Blog blog.px.dev. Checked 27 October 2026.
  4. Kubernetes Podcast from Google: Episode 150 - Pixie, with Zain Asgar and Ishan Mukherjee kubernetespodcast.com. Checked 27 October 2026.
  5. New Relic Signs Definitive Agreement to Acquire Pixie Labs, a Next Generation Machine Intelligence Observability Solution for Developers Using Kubernetes | New Relic newrelic.com. Checked 27 October 2026.
  6. Open sourcing Pixie under Apache 2.0 license | Pixie Labs Blog blog.px.dev. Checked 27 October 2026.
  7. [2203.05742] Bringing Source-Level Debugging Frameworks to Hardware Generators arxiv.org. Checked 27 October 2026.
  8. [2507.19635] Efficient and Scalable Agentic AI with Heterogeneous Systems arxiv.org. Checked 27 October 2026.
  9. [2511.13274] KForge: Program Synthesis for Diverse AI Hardware Accelerators arxiv.org. Checked 27 October 2026.
  10. [2606.02963] KForge: LLM-Driven Cross-Platform Kernel Generation for AI Accelerators arxiv.org. Checked 27 October 2026.
  11. Announcing Gimlet's Series B | Gimlet Blog gimletlabs.ai. Checked 27 October 2026.
  12. Scaling low-latency inference with Gimlet Cloud and Cerebras | Gimlet Blog gimletlabs.ai. Checked 27 October 2026.
  13. About Us | Gimlet Labs gimletlabs.ai. Checked 27 October 2026.
  14. Zain Asgar | Computer Science cs.stanford.edu. Checked 27 October 2026.