I’m an M.S. student in Computer Science at the University of Washington’s Paul G. Allen School of Computer Science & Engineering and a graduate student researcher at the Allen Institute for AI (Ai2). My research spans reinforcement learning for language models, multimodal foundation models, and robot learning.

At Ai2, I work on the AllenNLP scaling team, integrating multimodal capabilities into OLMo-core for mixture-of-experts language models. I’m also a member of the Social RL Lab, advised by Prof. Natasha Jaques, where I study multi-turn and self-play reinforcement learning for LLMs, and I work with Prof. Ranjay Krishna in the RAIVN Lab and Ai2’s PRIOR robotics team, where I was a core contributor to MolmoAct2. I have also been fortunate to be advised by Jason Ren and Jiafei Duan.

Before graduate school, I earned a B.S. in Computer Engineering with Distinction and a B.A. in Comparative History of Ideas from the University of Washington, where I wrote an honors thesis on hardware acceleration of neural networks with Prof. Scott Hauck in the ACME Lab. I have also interned as a software engineer on AWS SageMaker at Amazon, and I’ve served as Head TA for UW’s graduate and undergraduate machine learning courses.

Selected publications

• Demystifying Reinforcement Learning Post-Training of Language Models. D. Clay, et al. Under review at ICLR 2027. [arXiv]
• MolmoAct2: Action Reasoning Models for Real-world Deployment. H. Fang, J. Duan, D. Clay, et al. CoRL 2026 (Spotlight). [arXiv]
• Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning. M. Abdulhai, R. Cheng, D. Clay, et al. NeurIPS 2025. [arXiv]

See the publications page for the full list.