Daniel Fried

Daniel Fried is an Assistant Professor at the Language Technologies Institute within the School of Computer Science at Carnegie Mellon University. His research bridges natural language processing, pragmatics, and human-computer interaction, with a focus on making programming more communicative through language models. He previously held postdoctoral positions at FAIR Seattle and the University of Washington, earned a PhD from UC Berkeley, M.Phil from Cambridge, and B.S from the University of Arizona.

• University: Carnegie Mellon University
• Department: Language Technologies Institute
• Ranks: Assistant Professor

Research Interests: Daniel Fried's work centers on grounded interaction, where language models perform tasks in real-world contexts. His key domains include code generation (e.g., InCoder, StarCoder), multi-agent pragmatic modeling, and multimodal dialogue systems. He explores how agents can infer user intent through language, enhance code synthesis via reinforcement learning, and improve factuality in embeddings.

Recent Article Trends: His 2025 publications emphasize multi-agent coordination, programmatic skill induction, and execution-based code benchmarks. Work like RepoST and Agent Workflow Memory highlights scalable repository-level environments and memory mechanisms for agentic tasks. Papers such as CRScore and Dynamic Coalition Structure Detection reflect his focus on pragmatic evaluation and strategic language modeling.

• Scientific Awards: Okawa Research Grant (2023), Best Paper Award at Deep Learning for Code Workshop (2023), Best Paper Award at CVPR (2012)

Advising: He mentors PhD students at CMU, including Andre He, JY Koh, and Zora Wang. His alumni advisees, like Alex Xie, have contributed to machine learning and education projects.