Postdoctoral Associate

About this position

The DukeNLP group in the Department of Computer Science at Duke

University invites applications for a postdoctoral associate position

in natural language processing and machine learning. The group studies

how language models acquire, use, and communicate information, and how

these systems can be evaluated and deployed responsibly in

consequential settings. We are particularly interested in

understanding how errors arise and compound in extended or agentic

workflows, and how humans can efficiently assess and verify

AI-generated work. Possible research directions include developing

methods and evaluations for factual and robust language generation,

studying the behavior of models and agents over long-horizon tasks,

and measuring or reducing the human effort required to verify

AI-assisted outputs. Related work on the provenance, transparency, and

accountability of human-AI-produced artifacts is also welcome.

The specific research agenda will be developed jointly with the

successful candidate. The postdoctoral associate will be encouraged to

take intellectual ownership of one or more projects, develop new

research directions, publish at leading NLP and machine learning

venues, and release useful datasets, models, or evaluation resources.

The researcher will also collaborate with graduate and undergraduate

students and may participate in interdisciplinary projects with

researchers across Duke and other institutions. Applicants should

hold, or expect to complete before the start date, a PhD in computer

science, machine learning, natural language processing, or a closely

related field. Candidates should have:

• A demonstrated record of original research in NLP, machine learning, or a related area
• Experience conducting empirical research with large language models
• Strong programming and experimental skills
• The ability to formulate and pursue research questions independently
• Strong written and oral communication skills Experience with human-centered evaluation, reinforcement learning or post-training, agentic systems, or large-scale data collection is helpful but not required. We welcome candidates whose background complements the group’s existing expertise and who are interested in shaping a broader research program around reliable and accountable AI. The postdoctoral associate will be supervised by Professor Bhuwan Dhingra and will be part of the broader NLP and AI research community at Duke. More information is available at https://nlplab.cs.duke.edu/. The initial appointment is for one year, with the possibility of renewal for a second year contingent on satisfactory performance, continued funding, and mutual interest.