creators

Nathan Lambert - Interconnects, RLHF and open models

Nathan Lambert writes the Interconnects newsletter on RLHF, post-training and open models. He led post-training at Ai2 until June 2026 and wrote the RLHF book.

Nathan Lambert is a machine learning researcher who works on open language models. He took a PhD in electrical engineering and computer science at UC Berkeley (2017-2022) with a thesis on model-based reinforcement learning, after a BS in electrical and computer engineering at Cornell, and interned at Facebook AI and DeepMind. He was a research scientist and RLHF team lead at Hugging Face from 2022 to 2023. He joined the Allen Institute for AI (Ai2) in Seattle in 2023 and led its post-training work on the Olmo and Tulu models. His last day there was 2 June 2026; he has said he is building something new but has not named it, and his GitHub profile now lists Interconnects.ai as his organisation. Lambert writes Interconnects, a newsletter and podcast covering new models, the methods used to train them and the open-model ecosystem, and runs the ATOM Project, which argues for American investment in open models. Manning published his textbook Reinforcement Learning from Human Feedback in July 2026.

Specialization

RLHF, LLM post-training, open language models, reinforcement learning

Country

United States