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

Support our project (opens Buy Me a Coffee)
USASI

Kilian Lieret

AI research scientist at Meta Superintelligence Labs; lead author of the SWE-sweep bug-fixing benchmark

  • AI Research Scientist, Meta Superintelligence Labs (joined February 2026), Meta123

Kilian Lieret is an AI Research Scientist at Meta Superintelligence Labs, which Lieret joined in February 2026. Lieret is the first author of SWE-sweep, a Meta benchmark that asks agents to discover and repair as many bugs as they can in a real repository. Before Meta, Lieret was at Princeton University researching agentic AI for software engineering and graph neural networks for high energy physics, and co-authored SWE-agent and SWE-smith.13456

Career

At Princeton University, Lieret researched graph neural networks for particle tracking in high energy physics experiments, giving talks on object condensation pipelines in 2023, and then agentic AI for software engineering. Lieret joined Meta FAIR as an AI Research Scientist in February 2026, and Lieret's site now gives the role as AI Research Scientist at Meta Superintelligence Labs.1

Software engineering agents

Lieret co-authored SWE-agent (2024) with John Yang, Carlos E. Jimenez, Ofir Press and others, and SWE-smith (2025), which was accepted as a NeurIPS 2025 spotlight. Lieret's site also lists mini-swe-agent (July 2025), the CodeClash benchmark (November 2025) and ProgramBench (May 2026).561

SWE-sweep, released by Meta with Lieret as first author and John Yang as last author, asks agents to discover and repair bugs in real repositories without hints. Its scoring runs hidden tests per bug and scores a task as zero if the agent's patch breaks previously passing tests.34

Work

Career timeline

  1. May 2023

    Presented an object condensation pipeline for charged particle tracking at CHEP.1

  2. May 2024

    Co-authored the SWE-agent paper on agent-computer interfaces for software engineering.5

  3. Apr 2025

    Co-authored SWE-smith, on scaling data for software engineering agents.6

  4. Jul 2025

    Released mini-swe-agent, a minimal software engineering agent.1

  5. Feb 2026

    Joined Meta FAIR as an AI Research Scientist.1

  6. May 2026

    Released ProgramBench, a software engineering benchmark.1

Other documented roles

Roles the sources document but that are not confirmed as current.

  • Researcher in agentic AI for software engineering and graph neural networks for high energy physics, Princeton University1

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)

[![Kilian Lieret — Profile on USASI](https://unitedstatesofamericasuperintelligence.com/badges/profile.svg)](https://unitedstatesofamericasuperintelligence.com/people/kilian-lieret/)

HTML

<a href="https://unitedstatesofamericasuperintelligence.com/people/kilian-lieret/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/profile.svg" alt="Kilian Lieret — 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.
    Kilian Lieret — personal site (news, research) (external site: lieret.net)

    Kilian Lieret · Official page · accessed Oct 22, 2026

  2. 2.
    klieret (Kilian Lieret) on GitHub (external site: github.com)

    GitHub · Official page · accessed Oct 22, 2026

  3. 3.
  4. 4.
    facebookresearch/swe-sweep README (citation and evaluation) (external site: github.com)

    facebookresearch (GitHub repository) · Repository · accessed Oct 22, 2026

  5. 5.
  6. 6.

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