Jinyeop Song

sjinyeop@gmail.com / Google Scholar / GitHub / LinkedIn / X

I am an AI Researcher at Xaira Therapeutics, working in Bo Wang’s team on AI scientists (X-scientists) and virtual cell modeling. I received my PhD from the MIT Physics Department, advised by Prof. Jeff Gore (successfully defended in August 2026). I am a former MATS scholar and MIT-IBM Watson AI Lab research intern.

I work on the Science of AI and AI for Science. My recent work studies neural scaling laws, in-context learning, and LLM agents, while also building foundation-model and agentic systems for biology.

Selected publications

^† denotes co-first authors.

• Estimating the Empowerment of Language Model Agents. Jinyeop Song, Jeff Gore, Max Kleiman-Weiner. ICML 2026 / Paper
• The Blessing of Dimensionality in LLM Fine-tuning: A Variance-Curvature Perspective. Qiyao Liang, Jinyeop Song, Yizhou Liu, Jeff Gore, Ila R. Fiete, Risto Miikkulainen, Xin Qiu. arXiv 2026 / Paper
• Efficient and Transferable Agentic Knowledge Graph RAG via Reinforcement Learning. Jinyeop Song, Song Wang, Julian Shun, Yada Zhu. Under review 2026 / PDF
• Emergence and Effectiveness of Task Vectors in In-Context Learning: An Encoder Decoder Perspective. Jinyeop Song^†, Seungwook Han^†, Pulkit Agrawal, Jeff Gore. ICML 2025 Spotlight / Paper
• Reconciling Kaplan and Chinchilla Scaling Laws. Tim Pearce, Jinyeop Song. TMLR 2024 / Paper
• MethylGPT: a foundation model for the DNA methylome. Kejun Ying^†, Jinyeop Song^†, Haotian Cui^†, et al. bioRxiv 2024 / Paper

Experience

• AI Researcher, Xaira Therapeutics — South San Francisco, CA (Aug 2026 – Present)
• Research Intern, IBM Research – Watson AI Lab — Cambridge, MA (2025)
• Research Fellow, Machine Learning Alignment and Theory Scholars (MATS) — Berkeley, CA (2025)
• Graduate Research Assistant, MIT (PI: Prof. Jeff Gore) — Cambridge, MA (2021 – 2026)
• Research Intern, Samsung Electronics Research — Suwon, Korea (2020)
• Research Intern, Harvard–MIT Health Sciences and Technology — Boston, MA (2019)
• Undergraduate Research Assistant, KAIST — Daejeon, Korea (2014 – 2020)

Teaching and service

• Organizer, AI Journal Club, MIT (2024 – 2025)
• Teaching Assistant, "Complex Systems Biology", MIT (2023)
• Freshmen Tutoring Program, KAIST (2015 – 2018)

Awards and honors

• MATS Scholarship, 2025
• Mokam Research Scholarship, 2023
• Korean Physical Society Best Oral Presentation Award, 2020
• Asan Foundation Biomedical Science Scholarship, 2021-2025
• Excellent Graduate of KAIST, 2020
• KAIST Presidential Fellowship (KPF), 2016
• Korea Presidential Science Scholarship, 2014-2020

Projects

• Protein LLM: adapting open-source LLMs to protein modality via post-training.
• Autoresearch De Novo: LLM agents for inference-time protein design optimization.

See publications for a fuller list.