Andrii Zahorodnii

zaho [at] mit [dot] edu

About

I am a Member of Technical Staff at Merge Labs, working at the intersection of neuroscience and machine learning. I completed my undergraduate and master's studies at MIT in Computer Science and Brain and Cognitive Sciences, where my research focused on representation learning from large-scale human neural recordings. At MIT, I was advised by Prof. Ila Fiete and previously worked with Dr. Guangyu Robert Yang and Dr. Christopher Cueva. I co-founded the Ukraine Leadership and Technology Academy to equip young Ukrainians with the skills, knowledge, and confidence they need to rebuild Ukraine into a thriving, European nation. I won the Schwarzman Scholarship and am a Cerebras Fellow. I co-organize TEDxMIT. In 2025, I received the Glushko Prize for Outstanding Research; in 2026, I was elected to Full Membership in Sigma Xi, The Scientific Research Honor Society.

Projects & Publications

Two TESS transit signals for planet candidate TOI-790.03 from a SCI agent publication (top-row crop).
From a SCI agent publication: a second transit of planet candidate TOI-790.03. CC BY 4.0

SCI - Developing AI research taste through scientific discovery

[ website ] [ code ]

An open-source publication platform for AI agents and humans. Agents investigate open scientific data, publish signed, verifiable findings with code and data, and build on or challenge earlier work.

Noninvasive selective activation and suppression of the human brain with natural images

[ thesis ]

Deep neural networks trained on human intracranial recordings identify natural images that selectively drive or suppress activity at targeted brain sites. This master’s thesis provides a proof of concept for “writing” into the brain through sensory input, a potential route toward noninvasive brain-computer interfaces.

BT-bench visualization

Neuroprobe: Evaluating Intracranial Brain Responses to Naturalistic Stimuli

[ paper ] [ GitHub ] [ Leaderboard Website ]

Andrii Zahorodnii* · Christopher Wang* · Geeling Chau* · Bennett Stankovits · Charikleia Moraitaki · Eli Gross · Alexander Brady · Andrei Barbu · Boris Katz · Ila R. Fiete arXiv preprint (2026)

Macaque MST firing rates and cue-to-delay tuning inversion from Figure 5a

Overcoming sensory-memory interference in working memory circuits

[ paper ]

Andrii Zahorodnii · Diego Mendoza-Halliday · Julio C. Martinez-Trujillo · Ning Qian · Robert Desimone · Christopher J. Cueva under review at Nature Neuroscience (2025)

Project Jenkins: Monkey Thinks, Robot Moves... and Back?

[ project page ] [ code ]

Andrii Zahorodnii · Dima Yanovsky at MIT (2025)

Flappy Bird environment, lidar observations, actions, and world-state features from Figure 1

Improving World Models using Deep Supervision with Linear Probes

[ paper ]

Andrii Zahorodnii in ICLR 2025 Workshop on World Models, and on arXiv (2025)

Reviewer reliability model and comparison of simple averaging with Bayesian weighting from Figure 2

Paper Quality Assessment based on Individual Wisdom Metrics from Open Peer Review

[ paper ]

Andrii Zahorodnii · Jasper J.F. van den Bosch · Ian Charest · Christopher Summerfield · Ila R. Fiete on arXiv (2025)

The engagement of the cerebellum and basal ganglia enhances expertise in a sensorimotor adaptation task

[ paper ]

Joshua B Tan · Eli Müller · Andrii Zahorodnii · James M Shine in Imaging Neuroscience, Vol. 2, pp. 1-20, MIT Press (2024)

Writing

How to make Massive Open Online Courses work for Ukraine Leadership and Technology Academy (ULTA; 2023)

My journey of leadership: creating ULTA

This website template was inspired by that of Christopher Wang at MIT.