Finbarr Timbers Says He Is Leaving Ai2 After Open-Model Work
The AI researcher and investor, whose work spans reinforcement learning, game AI and generative models, praised Ai2's open-model push but did not name his next role.
By Ryan Merket · · updated
Why it matters
AI hiring and departures are now market signals. Finbarr's exit gives little away about his next move, but his emphasis on Ai2's infrastructure points to the compute race behind open model work.
Finbarr Timbers (@finbarrtimbers) said he is leaving Ai2, ending a stint he described as work on "the open model frontier" in a thread on X.
https://x.com/finbarrtimbers/status/2061824037261808037
"Ai2 is a fantastic institution," Timbers wrote. "I'm grateful for the opportunity to work there and push the open model frontier." He added that the team "will go on to do great things" and said he was "very excited for what they have planned," including what he described in parentheses as "GB300s."
Timbers describes himself as an AI researcher and investor working on reinforcement learning for LLMs. He previously worked at Midjourney and DeepMind, and writes about AI research, with stated interests in generative AI, large models, reinforcement learning and the future of technology.
His research record is concentrated in game-playing systems, multiagent reinforcement learning and software engineering. Listed work includes "Mastering the game of Stratego with model-free multiagent reinforcement learning," a 2022 Science paper cited 425 times; "OpenSpiel: A framework for reinforcement learning in games," a 2019 arXiv paper cited 411 times; and "Student of Games: A unified learning algorithm for both perfect and imperfect information games," a 2023 Science Advances paper cited 108 times. Other cited work includes "Detecting duplicate bug reports with software engineering domain knowledge," published in 2017 and cited 128 times, and "Computing approximate equilibria in sequential adversarial games by exploitability descent," a 2019 arXiv paper cited 103 times.
The post did not state where Timbers is going next or whether he is starting something new. That omission matters because talent movement has become one of the clearest signals in AI: model labs are competing on researchers, infrastructure access and the ability to ship credible open alternatives to closed frontier systems.
Timbers's short announcement also put compute at the center of the story. His aside about missing "the infra" and his reference to GB300s framed Ai2's next phase less as a personnel story than as a bet on the hardware needed to keep open models competitive.