Postdoctoral Associate

Join DukeNLP to study reliable language generation, model and agent behaviour over long tasks, and evaluation of AI-assisted work. Projects may address factual errors, transparency, provenance and the human effort needed to verify outputs. The researcher develops independent projects with Bhuwan Dhingra and collaborates with students. The initial appointment is one year, potentially renewable for a second subject to performance, funding and mutual interest.

Relevant PhD completed before the start date; original NLP or ML research, empirical LLM experience, strong programming, experimental and communication skills. Application materials include a cover letter and CV.