Independent project. Not a U.S. government website.

Support our project (opens Buy Me a Coffee)
USASI

Carlos Costa

IBM Distinguished Engineer and llm-d project lead for inference optimization research

  • Distinguished Engineer, IBM17
  • Project leadership and maintainer, llm-d12

Carlos Costa is a member of the project leadership of llm-d, the Kubernetes-native distributed inference serving stack, listed with IBM as employer, and one of the project's three initial maintainers, representing inference optimization research. llm-d's launch and release posts list Costa as a Distinguished Engineer at IBM.1237

llm-d

The llm-d project document names Costa, Clayton Coleman and Robert Shaw as the initial maintainers, with Costa representing inference optimization research. The current maintainers file lists Costa, with IBM as employer, in the project leadership with Abdullah Gharaibeh and Robert Shaw.21

Costa co-wrote "Announcing the llm-d community!" in May 2025, which presented llm-d as a modular serving stack using KV-cache aware routing and disaggregated serving with Kubernetes tooling. Costa also co-wrote the release posts for llm-d 0.2, 0.3, 0.4 and 0.5.34567

Work

  • llm-dOpen-source code

    Named one of llm-d's three initial project maintainers, representing inference optimization research, and co-author of the project's launch post and its v0.2 to v0.5 release announcements.21347

    Project page (external site: github.com)

Career timeline

  1. May 20, 2025

    Co-authored the post announcing the llm-d community.3

  2. Jul 29, 2025

    Co-authored the llm-d 0.2 release post on the project's first well-lit paths.4

  3. Oct 10, 2025

    Co-authored the llm-d 0.3 release post on wider hardware support.5

  4. Dec 2, 2025

    Co-authored the llm-d 0.4 release post on performance across accelerators.6

  5. Feb 4, 2026

    Co-authored the llm-d 0.5 release post, "Sustaining Performance at Scale".7

More sourced profiles at the same organization, in the same field, or from the same school.

Is this you?

Profiles cover public professional work only, and every line cites a source. If something is out of date, missing or wrong, send a correction with a link to an official page and we will update it. Suggest a correction

Profile on USASIAdd this badge to your website or GitHub profile README. It links back to this page. It is not an endorsement or a rating.

Markdown (for a README)

[![Carlos Costa — Profile on USASI](https://unitedstatesofamericasuperintelligence.com/badges/profile.svg)](https://unitedstatesofamericasuperintelligence.com/people/carlos-costa/)

HTML

<a href="https://unitedstatesofamericasuperintelligence.com/people/carlos-costa/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/profile.svg" alt="Carlos Costa — Profile on USASI" height="20"></a>

This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.

References · reviewed Oct 22, 2026

  1. 1.
    llm-d MAINTAINERS.md (external site: github.com)

    llm-d (GitHub repository) · Repository · accessed Oct 22, 2026

  2. 2.
    llm-d PROJECT.md (project ownership and initial maintainers) (external site: github.com)

    llm-d (GitHub repository) · Repository · accessed Oct 22, 2026

  3. 3.
    Announcing the llm-d community! (external site: llm-d.ai)

    llm-d · Announcement · published May 20, 2025 · accessed Oct 22, 2026

  4. 4.
    llm-d 0.2: Our first well-lit paths (external site: llm-d.ai)

    llm-d · Release notes · published Jul 29, 2025 · accessed Oct 22, 2026

  5. 5.
    llm-d 0.3: Wider Well-Lit Paths for Scalable Inference (external site: llm-d.ai)

    llm-d · Release notes · published Oct 10, 2025 · accessed Oct 22, 2026

  6. 6.
    llm-d 0.4: Achieve SOTA Performance Across Accelerators (external site: llm-d.ai)

    llm-d · Release notes · published Dec 2, 2025 · accessed Oct 22, 2026

  7. 7.
    llm-d 0.5: Sustaining Performance at Scale (external site: llm-d.ai)

    llm-d · Release notes · published Feb 4, 2026 · accessed Oct 22, 2026

Support Us

Help keep USASI useful.

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project