We invite applications for a two-year postdoctoral position in Natural Language Processing, starting March 1st, 2026, to help rethink LLM alignment with a long-term, goal-oriented perspective for social good. The position is funded by Villum Fonden (project “Aligning Multi-Agent Interactions for Sustainable Food Behavior (AMAI)”, grant VIL76720) and will be hosted by the CoAStaL group under the NLP section at DIKU. The project’s central idea is to move beyond preference-tuning and align agent behavior to measurable sustainability outcomes by coupling user-facing LLM agents with tool-using computational agents that estimate environmental impacts and predict adherence—initially in the domain of household food waste reduction.

Information on the department can be found at: https://di.ku.dk/



Our research
CoAStaL is a collective, open research culture that prizes diversity, friendliness, and life-work balance. Our team spans five continents and varied worldviews; we treat these differences as a source of learning and strength. Projects are open for everyone to join, ideas are shared freely, and we foster inclusion, engagement, trust, and respect—aiming to be not just a top research group but the friendliest in the top. We value ambition without ego-attachment: celebrate progress, learn from failures, and keep “shooting for the next deadline.” The result is a supportive environment with strong mentoring, shared infrastructure, and an international network of collaborators and visitors.

The AMAI project rethinks LLM alignment around long-horizon, outcome-oriented objectives for social good. Methodologically, we combine agentic LLMs with tool use, behavioral modeling, and field deployments to evaluate real impacts (e.g., sustained changes in household food practices) rather than short-term conversational rewards. A core interdisciplinary collaboration is with the Department of Food Science (UCPH), where we co-design and test recipe/meal-planning and persuasive communication interventions, integrate sensory and consumer research insights, and measure adherence and environmental outcomes in real-world settings.



Your job
The postdoc will conduct research on:

• Designing and training multi-agent LLM systems that coordinate tool calls (e.g., inventory parsing, footprint calculators) and generate persuasive, user-tailored narratives and recipes that promote sustainable choices.
• Developing and evaluating alignment objectives that optimize delayed, real-world outcomes (e.g., waste and CO₂ reductions) rather than short-term conversational satisfaction; exploring RL/RLHF, bandits, and distribution-level evaluation.
• Building and analyzing a study platform (mobile/web) to run a year-long field deployment with consenting participants, including data collection, adherence modeling, and ethics/privacy protocols.

Profile
We are looking for a highly motivated and enthusiastic scientist with the following competencies and experience:

Essential experience and skills:

• You have a PhD in NLP/ML/AI.
• You are highly experienced, as evidenced by first-author publications, in LLMs/agentic workflows and tool use; reinforcement learning or learning with delayed rewards; and evaluation/metrics for behavioral or distributional outcomes.
• You are committed to AI for social good, research integrity, and transparent, user-aligned systems.
• Proficient communication skills and ability to work in teams.
• Excellent English skills, written and spoken.

Desirable experience and skills:

• Experience with persuasive text generation/argumentation, recommendation, HCI/UX for AI systems, or field experiments.
• Familiarity with sustainability/food systems datasets or life-cycle/footprint calculators.
• Mobile/web app prototyping for research deployments.
• Open-science practices (data/model releases with safeguards and documentation).

Place of employment

The place of employment is at the Department of Computer Science (DIKU), University of Copenhagen. We offer creative and stimulating working conditions in dynamic and international research environment. Our research facilities include a high-performance GPU cluster for computational experiments.

Opportunities for short international research stays are planned to strengthen recipe modeling and computational argumentation aspects of the project.



Terms of employment
The average weekly working hours are 37 hours per week.

The position is a fixed-term position limited to a period of 2 years. The starting date is March 1st, 2026, or as soon as possible thereafter.

Salary, pension and other conditions of employment are set in accordance with the Agreement between the Ministry of Taxation and AC (Danish Confederation of Professional Associations) or other relevant organisation. Currently, the monthly salary starts at 39,200 DKK/approx. 5,200 EUR (November 2021 level). Depending on qualifications, a supplement may be negotiated. The employer will pay an additional 17.1 % to your pension fund.



Foreign and Danish applicants may be eligible for tax reductions, if they hold a PhD degree and have not lived in Denmark the last 10 years.



The position is covered by the Job Structure for Academic Staff at Universities 2020.



Questions
For further information please contact Daniel Hershcovich (dh@di.ku.dk), DIKU.

Foreign applicants may find this link useful: www.ism.ku.dk (International Staff Mobility).



Application procedure
Your online application must be submitted in English by clicking ‘Apply now’ below. Furthermore, your application must include the following documents/attachments – all in PDF format:

1. Motivated letter of application (max. one page).
2. CV incl. education, work/research experience, language skills and other skills relevant for the position.
3. A certified/signed copy of a) PhD certificate and b) Master of Science certificate. If the PhD is not completed, a written statement from the supervisor will do.
4. List of publications.

Deadline for applications: 19 November 2026, 23.59pm CET

We reserve the right not to consider material received after the deadline, and not to consider applications that do not live up to the abovementioned requirements.



The further process
After the expiry of the deadline for applications, the authorized recruitment manager selects applicants for assessment on the advice of the hiring committee. All applicants are then immediately notified whether their application has been passed for assessment by an unbiased assessor. Once the assessment work has been completed each applicant has the opportunity to comment on the part of the assessment that relates to the applicant him/herself.

You can read about the recruitment process at http://employment.ku.dk/faculty/recruitment-process/



The applicant will be assessed according to the Ministerial Order no. 242 of 13 March 2012 on the Appointment of Academic Staff at Universities.



Interviews are expected to be held in week 51.

The University of Copenhagen wish to reflect the diversity of society and encourage all qualified candidates to apply regardless of personal background.