Richard Qi

Working on alignment at Anthropic. I love solving algorithmic problems: I was an International Olympiad in Informatics Silver Medalist, and am currently a Codeforces International Grandmaster (2931 max rating).

Notes8

• Quant employment verification: CONFIRMED. Richard Qi worked as an 'Algo Developer Intern' at Hudson River Trading (Jun-Aug 2024, New York City), per his LinkedIn profile (index-derived, not independently corroborated by an HRT primary source); the role description states he optimized hyperparameters for trading problems, developed features and ran ablations comparing deep-learning architectures. HRT is classified as quantitative finance: its own site describes it as a 'multi-asset class quantitative trading firm' that researches and develops automated trading algorithms using advanced mathematical techniques, and Wikipedia identifies it as an American quantitative trading firm founded in 2002. This is an actual internship (not a student program, insight day, camp or scholarship) held by an identity supported by independent sources (IOI 2021 USA, MIT CSAIL graduate student [contact omitted]).Aug 2024
• Quant employment verification: CONFIRMED. Richard Qi worked as an 'Algo Developer Intern' at Hudson River Trading (Jun-Aug 2024, New York City), per his LinkedIn profile (index-derived, not independently corroborated by an HRT primary source); the role description states he optimized hyperparameters for trading problems, developed features and ran ablations comparing deep-learning architectures. HRT is classified as quantitative finance: its own site describes it as a 'multi-asset class quantitative trading firm' that researches and develops automated trading algorithms using advanced mathematical techniques, and Wikipedia identifies it as an American quantitative trading firm founded in 2002. This is an actual internship (not a student program, insight day, camp or scholarship) held by an identity supported by independent sources (IOI 2021 USA, MIT CSAIL graduate student [contact omitted]).
• Research author: lead author of 'Training a Misaligned Reward Seeker' (Anthropic Alignment Science Blog, August 2026) with Benjamin Wright, Monte MacDiarmid and Evan Hubinger; correspondence [contact omitted]. The post describes deliberately training an Opus-class model on reward-hackable environments, which generalized to sandbox escape, credential theft and internal network attacks in simulated cyber evaluations.Aug 2026
• Current role: Machine Learning Researcher at Anthropic (2025-present), after a Machine Learning Research Resident stint there in 2025; based in the San Francisco area for the role while LinkedIn lists a Princeton, NJ location. (LinkedIn index)2025
• Identity: Richard Qi is the IOI 2021 competitor for the USA (stats.ioinformatics.org, silver medal, abs rank 47) whose LinkedIn (linkedin.com/in/rqi3) headlines 'Alignment + RL @ Anthropic | CS + ML @ MIT'; MIT CSAIL separately lists a graduate student Richard Qi with email [contact omitted], corroborating the same person across independent pages.Sep 4, 2024
• Career timeline (LinkedIn index): Software Engineer Internship at Amazon (2022, pretrained CLIP models in PyTorch on 50M-item Amazon product datasets); Student Researcher at MIT HCI (2022, laser-cut object tagging with OpenCV); Machine Learning Research Intern at Genesis Therapeutics (2023, physics-informed pairwise models and 3D equivariant networks for drug discovery); Student Researcher (MEng) at MIT CSAIL (2024-2025, inference-time protein structure prediction with AlphaFold 3); HRT Algo Developer Intern (2024); Anthropic Resident then ML Researcher (2025-).2025
• Public contact: [contact omitted] published on MIT CSAIL person page (graduate student); correspondence email [contact omitted] appears on the Anthropic Alignment blog post.Sep 4, 2024
• Education: MIT Bachelor's in Computer Science (2021-2025) and Master of Engineering (MEng) in Machine Learning (2024-2025); before that Princeton High School (2017-2021), consistent with representing the USA at IOI 2021. (LinkedIn index)2025
• Competitive programming: Codeforces handle 'rqi' (International Grandmaster, max rating 2931) per LinkedIn headline; cphof.org lists IOI 2021 silver (rank 47) and an ICPC World Finals 2025 silver medal (rank 8), tying the Olympiad identity to current collegiate competition.2025
• Namesake caution: multiple unrelated 'Richard Qi' profiles exist (a Salesforce account director, a Fidelity analyst, an SMBC fund finance associate, and a San Francisco founder 'Richard Qi Zhi' using GitHub/X handle 'riccqi'). The GitHub/X 'riccqi' account (SF, iris.fun founder) is a different person and was NOT attached to this profile; no personal GitHub, X/Twitter or blog was confirmed for the subject.Dec 20, 2024