AI People Wiki

Builders & open source

Joel Schlosser

Listed with Meta on a June 2025 PyTorch blog post; GitHub profile lists Facebook as company, Meta

  • Listed with Meta on a June 2025 PyTorch blog post; GitHub profile lists Facebook as company, Meta34
  • Module maintainer, NN APIs (torch.nn) and Optimizers (torch.optim), PyTorch (open-source project)1

Joel Schlosser is listed on PyTorch's maintainers page as a module maintainer for the NN APIs (torch.nn) and the optimizers package (torch.optim), and as an emeritus maintainer of NestedTensor (torch.nested) and of the C++ API. The PyTorch CODEOWNERS file names Schlosser, with Alban Desmaison, as an owner of the torch/nn directory. Schlosser is listed with Meta as the first author of the June 2025 PyTorch blog post "Presenting Flux Fast: Making Flux go brrr on H100s," and Schlosser's GitHub profile lists Facebook as their company.1234

At a glance

Current roles
Listed with Meta on a June 2025 PyTorch blog post; GitHub profile lists Facebook as company +1 more
Companies
Meta, PyTorch
Sector
Company
Work listed
1
Sources cited
7
Profile reviewed

Open-source work

Schlosser maintains two of PyTorch's core modules, the neural-network APIs in torch.nn and the optimizers in torch.optim, and is named with Alban Desmaison as an owner of the torch/nn directory in the project's CODEOWNERS file. PyTorch's maintainers page also lists Schlosser as an emeritus maintainer of NestedTensor (torch.nested) and of the C++ API.12

Performance writing

Schlosser has co-written several PyTorch blog posts on speeding up generative models with native PyTorch features. A January 2024 post in the "Accelerating Generative AI with PyTorch" series used torch.compile and CUDA Graphs to speed up inference for FAIR's Seamless M4T-v2 translation model, and an April 2025 post described FlexAttention for inference, including a decoding kernel and paged attention.56

As first author of the June 2025 post "Presenting Flux Fast", written with Christian Puhrsch of Meta and Sayak Paul of Hugging Face, Schlosser showed how to make the Flux image-generation pipeline run faster on H100 GPUs.3

Work and contributions

  • PyTorch (torch.nn and torch.optim) Open-source code

    Listed as a maintainer of torch.nn and torch.optim and an emeritus maintainer of torch.nested and the C++ API; named in CODEOWNERS for /torch/nn/. The repository is under a BSD-style license.127

Timeline

  1. Co-authored the PyTorch blog post on accelerating Seamless M4T-v2 inference with native PyTorch.5
  2. Co-authored "FlexAttention Part II, FlexAttention for Inference" on the PyTorch blog.6
  3. Published "Presenting Flux Fast" on the PyTorch blog as first author.3

Where they work

  • Meta 106 people in the wiki
  • PyTorch 11 people in the wiki

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

7 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. PyTorch Governance | Maintainers (PyTorch 2.14 documentation)PyTorch · Documentation · published Jun 5, 2026 · accessed Oct 21, 2026
  2. pytorch/pytorch CODEOWNERSPyTorch (GitHub repository) · Repository · accessed Oct 21, 2026
  3. Presenting Flux Fast: Making Flux go brrr on H100sPyTorch · Announcement · published Jun 25, 2025 · accessed Oct 21, 2026
  4. jbschlosser (Joel Schlosser) on GitHubGitHub · Official page · accessed Oct 21, 2026
  5. Accelerating Generative AI with PyTorch IV: Seamless M4T, fast (PyTorch blog)PyTorch · Announcement · published Jan 23, 2024 · accessed Oct 21, 2026
  6. FlexAttention Part II: FlexAttention for Inference (PyTorch blog)PyTorch · Announcement · published Apr 30, 2025 · accessed Oct 21, 2026
  7. pytorch/pytorch LICENSE (BSD-style license)PyTorch (GitHub repository) · License · accessed Oct 21, 2026