Kasra Mazaheri

MIT master's student in AI focused on efficient machine learning (sparse attention, long-context LLMs, GPU kernels, diffusion models). Undergraduate CS at MIT; research assistant at MIT CSAIL and MIT RLE; teaching assistant in MIT EECS. Competitive-programming background as an IOI medalist for Iran (2019, 2020) and Codeforces 2900+; industry experience at Citadel (quantitative research), Hudson River Trading (algorithm engineering), The D. E. Shaw Group (software engineering) and Comcast (SWE/data science), per LinkedIn index records.

Notes5

• Quant employment verification: CONFIRMED. Kasra Mazaheri's LinkedIn index record shows a Quantitative Researcher role at Citadel (2025, Global Quantitative Strategies/GQS, San Francisco) alongside past Algorithm Engineer at Hudson River Trading (2024) and Software Engineer at The D. E. Shaw Group (2023). Citadel, HRT and D. E. Shaw are established quantitative hedge funds/market makers, so these are genuine quant-finance roles and not merely student programs. Caveat: this rests on the LinkedIn enrichment index and the caller-supplied leads, which are index/aggregator records rather than independent primary-source confirmation; no separate primary source (company page, press release, bylined announcement) corroborated the specific roles, and the caller's role dates (Citadel 2025-06..2025-09; HRT 2024-05..2024-08) were NOT independently confirmed - the index gives year-only dates. Identity is supported (MIT CS, IOI medalist, Codeforces KMAASZRAA).
• Career timeline (index-level, year-dated): Comcast SWE/Data Scientist (2022) -> D. E. Shaw Group Software Engineer (2023) -> Hudson River Trading Algorithm Engineer (2024) -> MIT CSAIL researcher (2024-2025, overlapping) -> Citadel Quantitative Researcher (2025), with MIT EECS TA (2022-2025) running in parallel throughout. Three successive quantitative-finance roles while still an MIT student.
• Personal site (last updated 2026) confirms he is currently a master's student at MIT researching efficient ML, framing him as a researcher rather than a full-time quant - which suggests the listed industry roles are internships/early-career stints rather than a settled finance career.2026
• Personal site schema.org metadata lists email [contact omitted] and profiles GitHub/KasraMazaheri, LinkedIn /in/kasra-mazaheri, Codeforces KMAASZRAA, and Google Scholar user=7uwtveQAAAAJ - the cross-links that support the identity.2026
• Research output at MIT Han Lab: co-author of 'Flash MoBA: Optimizing Mixture of Block Attention' (with Guangxuan Xiao, Junxian Guo, Song Han; arXiv 2511.11571, code at mit-han-lab/flash-moba), placing him in MIT's efficient-ML group.2025
• IOI record: competed for Iran at IOI 2019 and IOI 2020, earning one gold and one silver; Codeforces handle KMAASZRAA cross-linked from his own site and the IOI statistics page, tying the competitive-programming identity to the real name.2020
• Wrote and maintained educational Olympiad resources including GTOI, a freely available book on Computational Graph Theory for Olympiad, and taught CS at Farzanegan 1 Tehran and Shahid Soltani - a sustained teaching/access thread rather than a single gig.
• Google Scholar profile (user=7uwtveQAAAAJ) and GitHub (KasraMazaheri) are both cross-linked from his personal site, confirming those accounts belong to him.