Builder

Matej Vrzala

Matej Vrzala co-authored FrontierMath, a benchmark for mathematical reasoning in AI, and built interactive tools for Epoch AI's economic model and compute forecasts. In April 2020, he worked as a web developer at Metaculus and contributed to epidemicforecasting.org, a COVID-19 forecast created by Oxford's Future of Humanity Institute researchers and volunteer collaborators. He writes EA Slovakia's monthly newsletter, launched in September 2025; its October 2025 issue covered his application to a scientific panel on AI risks. In June 2026, he reported the group's plan for a civic association and meetings in Bratislava and online. As of October 2026, he contributes AI-safety and applied-ethics writing to GROWNi.

Timeline

  1. April 2020

    Web developer at Metaculus

    In April 2020, he works as a web developer at Metaculus and contributes to epidemicforecasting.org, a COVID-19 forecasting site created by researchers at Oxford's Future of Humanity Institute with volunteer collaborators.

  2. November 2024

    Co-authoring FrontierMath

    In November 2024, he co-authors FrontierMath with Elliot Glazer, Tamay Besiroglu, Jaime Sevilla and other co-authors. The benchmark contains hundreds of original, exceptionally challenging mathematics problems crafted and vetted by expert mathematicians, and state-of-the-art AI models solve under 2% of them.

  3. 2025

    The GATE model sandbox

    In 2025, he co-authors GATE: An Integrated Assessment Model for AI Automation with Ege Erdil, Andrei Potlogea and others. The model examines the economic impact of AI development, and he designs and implements its interactive sandbox.

  4. May 2025

    Compute-threshold projections

    In May 2025, with Edu Roldán, he implements the interactive tool and figures for an Epoch AI report on how many AI models will exceed training-compute thresholds. Its projections show 10 models above 1e26 FLOP by 2026 and more than 200 by 2030.

  5. September 2025

    EA Slovakia's monthly newsletter

    In September 2025, EA Slovakia (Efektívny Altruizmus Slovensko) launches a monthly newsletter, and he writes its first issue.

  6. October 2025

    Applying to an EU AI panel

    In October 2025, the second newsletter issue covers AI safety, and he writes that he has applied to a 60-member scientific panel tasked with alerting the EU AI Office to risks from large general-purpose AI models.

  7. June 2026

    EA Slovakia civic association

    In June 2026, he writes that EA Slovakia has planned a meeting in Bratislava with an online option and is seeking an unpaid co-founder for a civic association.

  8. As of October 2026

    GROWNi writer on AI safety

    By May 2026, his GROWNi article describes AI safety as an emerging field that seeks to prevent catastrophic problems. His September 2026 essay about Oxford philosophers discusses existential risk from technology, especially AI. He contributes articles to GROWNi on AI safety and applied ethics.

Sources and credits

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

  1. COVID-19 Forecast for India and Active infections estimate (fraction of population) publichealthupdate.com. Checked 21 October 2026.
  2. FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI arxiv.org. Checked 21 October 2026.
  3. GATE Model Playground epoch.ai. Checked 21 October 2026.
  4. How many AI models will exceed compute thresholds? epoch.ai. Checked 21 October 2026.
  5. Prvé číslo skea.substack.com. Checked 21 October 2026.
  6. Druhé číslo skea.substack.com. Checked 21 October 2026.
  7. Ako to vyzerá s AI? growni.sk. Checked 21 October 2026.
  8. Ako efektívne pomáhať zvieratám? skea.substack.com. Checked 21 October 2026.
  9. Štyria Oxfordský myslitelia growni.sk. Checked 21 October 2026.