Arvind Satyanarayan

The Massachusetts Institute of Technology (MIT) is seeking a postdoctoral researcher in Human-AI Interaction, with a focus on reinforcement learning and ethnographic rewards. The position is part of a research effort to develop AI agents that promote human metacognition and sociality, using reinforcement learning with ethnographic rewards (RLER). This involves combining multi-turn RL with the observation of workplace skill development to derive reward functions for complex behaviors, such as knowing when to step back or prompting reasoning articulation. The successful candidate will work on designing novel interfaces, including productive friction, contrastive explanations, and thought-trace visualizations, and will collaborate with an interdisciplinary team spanning HCI, machine learning, anthropology, and sociology. The postdoc will work closely with Professors Arvind Satyanarayan, Mitchell L. Gordon, Zana Buçinca, Susan Silbey, and Graham Jones, and will be part of MIT's Social and Ethical Responsibilities of Computing (SERC) program. Applicants must hold a PhD in Computer Science or a closely related field, with hands-on experience in training or fine-tuning large language models. Candidates with interests in multi-turn RL, human-AI interaction, or AI alignment are strongly encouraged to apply. The position requires the ability to advance research independently and a willingness to engage in interdisciplinary collaboration. The initial appointment is for one year, starting September 2026, with the possibility of renewal for a second year. Applications are considered on a rolling basis until May 1, 2026, and early applications are strongly preferred. For more information and to apply, please see the provided Google Form link. This is a paid postdoctoral position; specific funding details are not disclosed in the announcement.