Daniel Fried

Daniel Fried studies how people and AI systems communicate and collaborate on grounded tasks. He earned degrees at Arizona, Cambridge and Berkeley, where his doctoral work focused on grounded pragmatic communication; he later coauthored research on instruction following and InCoder, a code infilling model. After postdoctoral work at Washington and a visiting research role at Facebook AI Research, he joined Carnegie Mellon and worked on code editing at GitHub Next. He received an Okawa Research Grant in 2023, an NSF CAREER Award in 2026, and was named a 2026 AI2050 Early Career Fellow. As of October 2026, he is an assistant professor at CMU and a part-time research scientist at Meta, where he works on LLM-based coding agents.

Now at: Carnegie Mellon University and Meta
Works: 49
Badges: Bridge

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

2010–2021

Degrees at Arizona, Cambridge and Berkeley

From 2010 to 2014, he earns a B.S. in computer science, mathematics and information science at Arizona, with thesis adviser Mihai Surdeanu. From 2014 to 2015, he earns an M.Phil. at Cambridge's Computer Laboratory, advised by Stephen Clark. From 2015 to 2021, he completes a Ph.D. in Berkeley's NLP Group and Berkeley AI Research Lab under Dan Klein, with a thesis titled “Learning Grounded Pragmatic Communication.”
2018

Pragmatic instruction models

In 2018, he coauthors “Unified Pragmatic Models for Generating and Following Instructions” with Jacob Andreas and Dan Klein. Its evaluation shows that pragmatic inference improves models that interpret human instructions and generate instructions people understand.
2021–2022

Postdoctoral work at Washington

From 2021 to 2022, he works as a postdoc at the University of Washington with Luke Zettlemoyer and as a visiting researcher at Facebook AI Research with Mike Lewis.
2022

The InCoder code model

In 2022, he coauthors InCoder, a generative model that performs code synthesis and editing through infilling. The model generates code using context on both sides of missing spans.
August 2022

Joining CMU's Language Technologies Institute

In August 2022, he joins Carnegie Mellon's Language Technologies Institute as an assistant professor. At LARC, he leads work on agents that learn from people and environments, communicate with each other, and reason about real-world tasks.
September 2023–2024

Code editing at GitHub Next

In September 2023, he receives an Okawa Research Grant for work on Pragmatic Language Interfaces. From 2023 to 2024, he consults for GitHub Next on LLM-based models for code editing.
2026

NSF and AI2050 honors

In 2026, he receives an NSF CAREER Award for research that improves interaction and communication between people and AI systems. Schmidt Sciences names him an AI2050 Early Career Fellow for research on trustworthy communication among teams of AI agents.
As of October 2026

CMU and Meta research roles

As of October 2026, he is an assistant professor in CMU's Language Technologies Institute and a Research Scientist at Meta, where he spends 20% of his time on LLM-based coding agents. He leads LARC, whose methods help agents learn from people, environments and one another, communicate effectively, and plan and reason about the world.