I am a Research Scientist at Sakana AI, where I work on AI systems for financial applications and research methods that bridge advances in machine learning with real-world deployment. Previously, I earned a Ph.D. in Data Science from the NYU Center for Data Science under the supervision of Kyunghyun Cho and Krzysztof Geras. My doctoral research focused on identifying and mitigating spurious correlations in real-world datasets, particularly in medical and biological applications.

Email: taro[at]sakana[dot]ai

News
Jan. 26, 2026: “EDINET-Bench: Evaluating LLMs on Complex Financial Tasks using Japanese Financial Statements” was accepted to ICLR 2026.

Mar. 17, 2025: I defended my Ph.D. at the NYU Center for Data Science.

Mar. 3, 2025: I joined Sakana AI as a Research Scientist on the Applied Team.

Sep. 25, 2024: “Jointly modeling inter- & intra-modality dependencies for multi-modal learning” was accepted to NeurIPS 2024.

Apr. 22, 2024: I am interning with the Regev lab at Genentech to work on causal representation learning.

Aug. 29, 2023: “Detecting incidental correlation in multimodal learning via latent variable modeling” was accepted to TMLR.

May 8, 2023: I am interning with the Prescient Design group at Genentech to work on causal representation learning.

Sep. 14, 2022: “Generative multitask learning mitigates target-causing confounding” was accepted to NeurIPS 2022.

Education
Ph.D. Machine Learning (4.0 GPA), NYU Center for Data Science, 2020 - 2025

M.Sc. Artificial Intelligence (4.0 GPA), University of Edinburgh, 2018 - 2019

B.A. Mathematics, Northwestern University, 2007 - 2009