Blätter-Navigation

Offer 1 out of 152 from 02/04/26, 10:24

logo

Wei­zen­baum-Insti­tut e.V.

Founded in 2017, the Weizenbaum Institute researches the effects of advancing digitalisation on our society. With its recommendations for action, it helps to ensure that the digital transformation is sustainable, self-determined and responsible. The Weizenbaum Institute is supported by a network of seven partners, including Freie Universität Berlin, Humboldt-Universität zu Berlin, Technische Universität Berlin, Berlin University of the Arts, the University of Potsdam, the Fraunhofer Institute for Open Communication Systems (FOKUS) and the Social Science Research Centre Berlin (WZB). The institute is financed by the Ministry of research, technology and space travel (BMFTR) and the state of Berlin. It is located in Berlin.

For our research project "Evaluating GenAI Evaluations: Epistemics, Stakeholders, Technical Infrastructure and Policy" at the Weizenbaum-Institut e.V. we are looking at the earliest possible date for a

Evaluations Engineering Student Assistant (m/f/d)

English

with 23,08 percent of regular working hours (9 hours per week). The position is initially for 4 months with the option of extension.

Work content
You will work with David Hartmann, Jan Batzner, Anne K. Krüger, Rainer Rehak, LK Seiling, and Angelie Kraft across six Weizenbaum research groups. Your task will focus on furthering a shared schema and crowdsourced database to create a common language for reporting and comparing evaluation results across frameworks — as part of the broader evaluations community.

Tasks:

\ Contributing to the development of shared evaluation schemas (eval.schema.json, instance_level.schema.json) and infrastructure for standardized reporting of GenAI evaluation results
\ Developing and maintaining Python-based tools, pipelines, and conversion scripts for evaluation data (e.g., from lm-eval-harness, Inspect AI, HELM)
\ Working with the Hugging Face stack to integrate evaluation infrastructure
\ Benchmarking storage formats and validation pipelines for instance-level evaluation data
\ Documenting findings and contributing to reports and publications
\ Participating in workshops, colloquia, and events at the Weizenbaum Institute

Requirements:

\ Enrolled in a Computer Science, Data Science, or related degree program (we consider all majors with demonstrable coding/development experience)
\ Solid Python skills
\ Experience working with the Hugging Face stack
\ Familiarity with JSON schemas, Pydantic, and/or database fundamentals
\ Interest in AI evaluation, benchmarking, and reproducibility
\ Experience with backend design, GitHub Actions, or CI/CD pipelines is a plus
\ Strong English skills (written and spoken); German is not required
\ Self-organized, team-oriented, and eager to dive into new topics
\ An interdisciplinary mindset

What we offer:

  • flexible working hours and the option of working in parts remotely (in consultation with team lead)
  • remuneration analogue to TV Stud III
  • insight into current research and everyday scientific work
  • close cooperation with the scientific staff
  • inspiring academic and international working environment in the heart of Berlin (near Bahnhof Zoo) with broad, interdisciplinary networks in research, media and politics

How to apply:

Severely disabled applicants with equal qualifications will be given preference. We value diversity and welcome all applications, regardless of gender, nationality, ethnic or social origin, religion, disability, age, or sexual orientation. The Weizenbaum Institute expressly encourages women* and people with a migration background to apply.

You can submit your application, consisting of a letter of motivation (max. 1 page), CV (max. 2 pages), relevant transcripts, your current proof of matriculation and your academic certificates and grades, in our application portal by 30th April 2026. If you have any questions, please feel free to contact Maite Vöhl from our HR team (personal[at]weizenbaum-institut.de) at any time. We look forward to receiving your documents.