Zhezheng Luo

Zhezheng Luo is a Member of Technical Staff at OpenAI in the San Francisco Bay Area, previously a Quantitative Research Analyst at Citadel Securities focused on options market-making strategy research (2023-2025). Representing the United States of America, he won a gold medal at IOI 2017, ranked 3rd overall. He holds an MEng in Computer Science (2021-2023) and a Bachelor's degree in Electrical Engineering & Computer Science and Mathematics (2017-2021) from MIT, where he did neuro-symbolic machine-learning research (TOQ-Nets, rational subgoals) and was a SuperUROP scholar supervised by Leslie P. Kaelbling.

Notes10

• Quant employment verification: CONFIRMED. He held a genuine past role as a Quantitative Research Analyst at Citadel Securities (2023-2025), explicitly described as 'Options Market-Making Strategy Research'. Citadel Securities is a wholesale market maker / electronic trading firm (equities, options and fixed income), so the role is a real quantitative-finance position - not a student program, camp, scholarship, insight day or competition. Identity is supported: the same person is the IOI 2017 USA contestant whose Codeforces handle matches his GitHub, and the MIT/OpenAI profile aligns with his MIT degrees. Caveat: the specific title and dates are index-derived (LinkedIn enrichment); no independent primary firm page re-confirming them was located, so treat title/dates as index-only while the employment substance is supported.2023
• Current role: Member of Technical Staff at OpenAI, San Francisco Bay Area, held since 2025. Source is LinkedIn-derived enrichment (index), not an independent primary confirmation.2025
• Career background in quantitative finance: Quantitative Research Analyst at Citadel Securities (Miami), 2023-2025, described as 'Options Market-Making Strategy Research' - a genuine options market-making quant-research role. Index-derived (LinkedIn enrichment).2023
• Identity anchor: IOI 2017 contestant 'Zhezheng Luo' for the United States of America is the same person as the MIT/OpenAI profile - the IOI stats page links his Codeforces handle 'c_sunshine' (C_SUNSHINE), which is also the GitHub account attributed to this person.2017
• Transition point: Citadel Securities ends 2025 and OpenAI begins 2025, consistent with a single job change rather than overlapping roles - the shared 2025 boundary reflects one moving from the other, not parallel employment.2025
• Education: MIT MEng Computer Science (2021-2023) and MIT BS Electrical Engineering & Computer Science and Mathematics (2017-2021) - he went from undergrad into the Citadel Securities quant-research role without a separate degree gap.2023
• IOI 2017 gold medal, ranked 3rd overall (466.27/600 points), representing the United States; the IOI stats record lists him as the USA contestant.Aug 4, 2017
• Competitive-programming record beyond IOI: silver medal at Google Hash Code 2020 (team 'simplicissimus', 2nd) and a bronze medal / North America Champion at the ICPC 2020 World Finals (held Moscow 2021) with the MIT team (Benjamin Qi, Spencer Compton), placing his peak competition activity in his MIT years.Oct 5, 2021
• ML research at MIT focused on neuro-symbolic relational models for action recognition and planning; his releases include TOQ-Nets (Temporal and Object Quantification Networks) and a rational-subgoals planning model, published as PyTorch repos under his C-SUNSHINE GitHub account.2021
• Academic footprint on Google Scholar: 42 citations and h-index 4 (since 2021), verified email at openai.com, labeled interest 'Machine Learning'; the openai.com Scholar email is index/Scholar-provided, not an independently published contact.2025
• MIT SuperUROP Undergraduate Research and Innovation Scholar (2020-2021, EECS), supervised by Leslie P. Kaelbling, on 'Action Grounding with Primitive Predicates and Logic Inference' - a selective research program, distinct from industry employment.2020
• Public preprint profile (alphaXiv) repeats the same four research threads (edge matching/inequalities, hypergraph neural network expressiveness, rational subgoals, TOQ-Nets) and lists OpenAI and MIT SuperUROP as experience - an aggregator repeating index/self-reported data, not independent primary confirmation.2025