PeopleofSuperintelligence

Researcher

Ahmad Beirami

Ahmad Beirami is an AI researcher whose work spans game-playing agents, dialogue systems and language-model alignment. He won a bronze medal in Iran's Mathematics Olympiad in 2002 and received Georgia Tech's Sigma Xi Best PhD Thesis Award in 2015; his academic training included PhD and postdoctoral work at MIT, Harvard, Duke and Georgia Tech. At Electronic Arts, he applied reinforcement learning to game playtesting and coauthored a 2020 paper on intelligent game-development agents; at Meta AI, he helped prototype CAIRaoke and coauthored CheckDST. At Google, he coauthored controlled-decoding research in 2024; in 2025, a paper he coauthored won ICLR's Outstanding Paper Award, and he coauthored theory on best-of-n alignment. As of October 2026, he is co-founder and CEO of Fidian, which checks every AI agent decision against a team's specification during development and production.

Now at
Fidian
Badges
Links
Google Scholar, X and LinkedIn

Timeline

  1. 2002

    Iranian Mathematics Olympiad bronze

    He wins a bronze medal in Iran's Mathematics Olympiad in 2002.

  2. 2015

    A Sigma Xi thesis award

    He receives Georgia Tech's Sigma Xi Best PhD Thesis Award in 2015.

  3. By 2018

    PhD and postdoctoral training

    His academic training includes PhD and postdoctoral work at MIT, Harvard, Duke and Georgia Tech. Before moving to industry in 2018, he holds a joint postdoctoral fellow position at Harvard and MIT.

  4. 2019–2020

    AI agents for game testing

    EA's Apex Legends credits list him among the company's data scientists in 2019. His Electronic Arts work applies reinforcement learning to game playtesting, accelerating pre-release bug discovery and balance tuning. In 2020, he coauthors “Winning Is Not Everything: Enhancing Game Development With Intelligent Agents,” published in IEEE Transactions on Games.

  5. 2022

    Dialogue diagnostics and an assistant

    He coauthors the 2022 EMNLP Findings paper Know Thy Strengths, which presents CheckDST, a toolkit of robustness tests and failure-mode analytics for dialogue state tracking. At Meta AI, he leads the cross-org effort to prototype CAIRaoke, Meta's first tool-using digital assistant.

  6. 2024

    Controlled decoding at ICML

    At Google, he coauthors Controlled Decoding from Language Models, published at ICML in 2024. Its prefix scorer guides generation from a frozen base model at inference time.

  7. 2025

    A safety award and best-of-n theory

    In 2025, his coauthored paper on safety-alignment depth wins an ICLR Outstanding Paper Award. It links shallow alignment to adversarial suffix, prefilling, decoding-parameter and fine-tuning vulnerabilities in language models. At ICML 2025, he coauthors Theoretical guarantees on the best-of-n alignment policy, which shows a commonly used expression for KL divergence is an upper bound on its actual value and studies how tight the bound is.

  8. As of October 2026

    Co-founder and CEO

    As of October 2026, he is co-founder and CEO of Fidian. The company checks every AI agent decision against a team's specification during development and production.

Career

  1. NowFidianCo-founder and CEO
  2. 2024Google
  3. 2022Meta
  4. 2019Electronic ArtsData Scientist
  5. Massachusetts Institute of TechnologyPostdoctoral Fellow
  6. Harvard UniversityPostdoctoral Fellow

Education

  1. Massachusetts Institute of Technology
  2. Harvard University
  3. Duke University
  4. Georgia Institute of Technology

From the biography.

  1. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Ahmad Beirami. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

Colleagues

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

Sources and credits

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

  1. Ahmad Beirami beirami.github.io. Checked 21 October 2026.
  2. Talk by Ahmad Beirami (Google Research) – USC Center on AI Foundations for Science (AIF4S) sites.usc.edu. Checked 21 October 2026.
  3. EA Credits careers.ea.com. Checked 21 October 2026.
  4. EA Credits ea.com. Checked 21 October 2026.
  5. Ahmad Beirami | Academic Engagements beirami.github.io. Checked 21 October 2026.
  6. Know Thy Strengths: Comprehensive Dialogue State Tracking Diagnostics aclanthology.org. Checked 21 October 2026.
  7. Controlled Decoding from Language Models proceedings.mlr.press. Checked 21 October 2026.
  8. Announcing the Outstanding Paper Awards at ICLR 2025 blog.iclr.cc. Checked 21 October 2026.
  9. Safety Alignment Should Be Made More Than Just a Few Tokens Deep arxiv.org. Checked 21 October 2026.
  10. Theoretical guarantees on the best-of-n alignment policy proceedings.mlr.press. Checked 21 October 2026.
  11. Verification and governance for AI agents | Fidian fidian.ai. Checked 21 October 2026.
  12. Stealth Startup Spy #364 - by Drake Dukes stealthstartupspy.substack.com. Checked 21 October 2026.