AutumnTHE OLYMPIAD COLLECTION
Menu
RESEARCHED WITH AUTUMN
J(

John (Zhiyao) Ma

Member of Technical Staff at Anthropic; formerly quantitative researcher at Millennium and Two Sigma
Anthropic · New York, United States · engineer
Silver Medal For Canada At IMO 2013Quantitative Researcher At Two Sigma And Millennium (2018-2024)Member Of Technical Staff On Anthropic's Pretraining Team (2024-present)Author Of Mamba-minimal, A Widely Used Single-file PyTorch Mamba Implementation

Engineer and former quantitative researcher. MIT (BS + MEng, CS/HCI, 2013-2018); IMO 2013 silver medalist for Canada. Quantitative researcher at Two Sigma (2018-2021) and Millennium Management (2022-2024), both systematic/multi-strategy quantitative investment firms. Since 2024 a Member of Technical Staff on the Pretraining team at Anthropic in New York and San Francisco. Author of the open-source mamba-minimal implementation of the Mamba SSM architecture.

John (Zhiyao) Ma is a Canadian computer scientist and quantitative researcher, now a Member of Technical Staff on the Pretraining team at Anthropic in New York and San Francisco. He earned a B.S. and a Master of Engineering in Computer Science / HCI from MIT (2013-2018), won a silver medal representing Canada at the 2013 International Mathematical Olympiad, and worked as a quantitative researcher at Two Sigma and Millennium from 2018 to 2024. He authored mamba-minimal, a widely used single-file PyTorch implementation of the Mamba state-space model.

Details
LocationNew York, United States
Company siteanthropic.com
UniversityMassachusetts Institute of Technology
Computer ScienceHuman-Computer InteractionMachine LearningPyTorchState-space modelsQuantitative Research
Notes
  • Quant employment verification: CONFIRMED. Ma's own site states 'Quantitative researcher at Two Sigma and Millennium (2018 - 2024)' - an actual employment role, not a student program, insight day, or offer - and identity is corroborated by the same site linking his GitHub, LinkedIn and IMO background. Employer classification: Two Sigma is a systematic/quantitative investment manager and hedge fund; Millennium Management is a global multi-strategy hedge fund; both genuinely operate in quantitative finance, so these roles qualify.
  • Landing page: 'Member of Technical Staff at Anthropic on the Pretraining team, living in New York & SF' - first-person current employer and team.
  • Self-reported career timeline: Member of Technical Staff at Anthropic (2024-present); quantitative researcher at Two Sigma and Millennium (2018-2024); BS and Masters from MIT in Computer Science/HCI (2013-2018); IMO silver medal (2013).
  • Authored mamba-minimal, a single-file PyTorch implementation of the Mamba state-space architecture; announced on X (20 Dec 2023) drawing ~840 likes and 106 reposts, with third-party reimplementations and directory listings confirming its reach.
  • Silver medal representing Canada at the 2013 IMO (rank 101, 27 points; official results table lists Canada, award S).
  • His own summary collapses Two Sigma and Millennium into '2018-2024' while the LinkedIn index dates them 2018-2021 and 2022-2024; there is an ~2021-2022 transition not documented at month level, so exact dates are left unfilled rather than inferred.
  • Personal site links a LinkedIn (john-ma-92442138) that differs from the johnma2006 LinkedIn handle in the caller lead and the enrichment index - possibly an old vs current account; neither is independently confirmed as his sole profile, so both the caller handle and this one are treated as unverified.
  • X account @johnma2006 (293 followers) lists bio '@ Anthropic', corroborating the Anthropic affiliation; the account is sparse, its most substantive post being the Mamba announcement.
  • Enrichment index (local index, not independent primary confirmation) lists title 'Member of Technical Staff', company masked as 'A' in New York (2024-), plus Quantitative Researcher at Millennium and Two Sigma; consistent with his personal site but the company label is redacted in the index.

Competition record

Canada · IMO

Related collections

Explore the collection