Ziqian Zhong

Ziqian Zhong is a PhD student at Carnegie Mellon University advised by Aditi Raghunathan, working on AI safety and interpretability (behavior/generalization, scalable interpretability, multi-agent safety). Previously studied computer science and math at MIT, working with Jacob Andreas and Max Tegmark, and worked as a research scientist at Pika, where he led development of Pika 2.0. IOI 2019 contestant representing China.

Notes3

• Quant employment verification: CONFIRMED. Ziqian Zhong held a paid Algo Developer Internship at Hudson River Trading (HRT), 05/2023-08/2023. Evidence: his own resume states 'Algo Developer Intern, Hudson River Trading, 05/2023 - 08/2023. Conducted market and algorithmic research; project featured in intern spotlights' (fjzzq2002.github.io/Resume-Academia.pdf); this is corroborated by HRT's own blog, 'Intern Spotlight: 2023 HRT AI Labs Summer Projects' (published 2023-09-21), bylined 'Jesse Chan, Ziqian Zhong & Emma Yang', which describes HAIL (HRT AI Labs) Algo Developer interns applying AI to trading problems. Employer classification: HRT operates in quantitative/systematic algorithmic trading (HRT's own machine-learning page: 'HRT AI Labs ... integrating cutting-edge AI research and infrastructure into our trading strategies'), so this is a genuine quantitative-trading role, not a company with only 'quant' in its name. The role was an actual internship (work), not a competition, insight day, camp, scholarship, or incoming offer. Identity is supported: the same resume (MIT CS+Math, IOI/Codeforces handle fjzzq2002) matches the LinkedIn identity lead by name, education and the HRT role.
• Identity resolution: the IOI 2019 China contestant Ziqian Zhong uses Codeforces handle fjzzq2002 and the personal site fjzzq2002.github.io; his own resume (MIT BS in CS & Mathematics 08/2020-06/2024; CMU PhD 01/2025-present) matches the LinkedIn identity lead linkedin.com/in/ziqian-zhong-9723a9146 (same MIT education and HRT Algo Developer Intern role), so the LinkedIn employment timeline corroborates the primary-source resume rather than standing alone.
• HRT internship detail: at HRT he did 'market and algorithmic research' on the HRT AI Labs (HAIL) team, and his project was featured in the 2023 HRT AI Labs intern spotlight, indicating the work was selected as representative of HAIL's AI-for-trading research rather than routine intern work.Sep 21, 2023
• Ongoing role: PhD in Computer Science at Carnegie Mellon University since 01/2025, advised by Aditi Raghunathan, researching AI safety and interpretability; the CMU PhD is his current position as of 2026, superseding the summer-2026 Transluce internship which ended 08/2026.
• Industry research role: Research Scientist at Pika Labs / Mellis Inc. 06/2024-01/2025, developed large-scale video generation models, contributed to Pika 1.5 and led the release of Pika 2.0 - a senior individual-contributor role taken immediately after his 2024 MIT graduation and before starting the CMU PhD.
• Career timeline (from his resume): Theoretical CS undergraduate researcher at MIT 10/2021-05/2022 (team guided by Virginia Vassilevska Williams, graph theory/combinatorics); Deep Learning undergraduate researcher at MIT 08/2022-06/2024 (with Neil Thompson, Jacob Andreas, Max Tegmark, Ziming Liu); showed a transition from pure theory into deep-learning interpretability during his undergraduate years.
• Competition identity: represented China at IOI 2019; his research and intern-work pages consistently use the Codeforces handle fjzzq2002, which is the public link tying the contest identity to the MIT/CMU researcher.
• Research Intern at Transluce 05/2026-08/2026 (bounded summer range, ended), publishing research on situation awareness with Cassidy Laidlaw and Jacob Steinhardt.
• Education: B.S. in Computer Science & Mathematics, MIT, 08/2020-06/2024, GPA 5.0/5.0; coursework included Quantitative Methods for NLP, Multi-agent Communication, Machine Learning, Advanced Complexity Theory and Fundamentals of Statistics.
• Invited talk at the Jane Street Research Symposium 2026 ('How can we stop reward seekers?') - this is a speaking appearance, not employment; Jane Street is a quantitative trading firm but the item does not establish any role at it.
• Other invited talks: NeurIPS 2023 Oral 'Two Stories in Mechanistic Explanation of Neural Networks' (with Ziming Liu), SODA 2023 'New Approach for Unbounded SubsetSum', MIT AI Alignment 2025 'Let's talk about reward hacking', and CMU AI Safety Initiative 2026 'Can we fix reward hacking?'.
• Publication record includes NeurIPS 2024 'Algorithmic Capabilities of Random Transformers', NeurIPS 2023 Oral 'The Clock and the Pizza', ICLR 2026 'ImpossibleBench' and 'Watch the Weights', and COLM 2026 'Pando'.