DIPF | Leibniz Institute for Research and Information in Education contributes to addressing challenges in education through empirical research, digital infrastructure and knowledge transfer. At its locations in Frankfurt am Main and Berlin, DIPF develops and documents knowledge about education and thus supports science, politics and practice.

The Information Center Education (IZB) department is looking for the following position in the Educational Technologies division, **starting April 1, 2026**:

# **A research assistant for a PhD**

Full-time, temporary from 01.04.2026 – 31.12.2029, salary according to EG 13 of the collective agreement for the public service of the state of Hesse (TV-H).

The place of employment is Frankfurt am Main.

The project Z03 „Educational Science Methodologies of Participation Challenged by Digital Technologies“ [in:just:MetaPART] is part of the DFG Collaborative Research Center 1750 [in:just] “Inclusion - Recognition - Justice. Participation (Teilnahme) and Belonging (Teilhabe) in Processes of Growing Up”, which is coordinated by the Goethe University Frankfurt. Project Z03 develops educational methodologies for participation using digital technologies. It combines critical social research with AI-supported text analysis and network analysis to examine processes of recognition, inclusion and justice in educational contexts. Z03 establishes a methodological infrastructure that integrates algorithmic procedures with qualitative research. It forms the methodological center of the SFB [in:just] and contributes to the development of theory and methodology through cross-project data analysis.

### Your tasks

- Development, training and calibration of an educational AI agent for automated annotation, coding and labeling support for transcribed interviews and other qualitative text data.
- Operationalization and implementation of complex coding schemes (e.g., participation & belonging, inclusion, recognition, justice) in NLP pipelines: from guidelines/definitions to label formats to model and evaluation design.
- Setting up and maintaining annotation workflows (e.g., with INCEpTION/WebAnno or similar): project setup, schema management, guideline versioning, quality assurance, export/import, consistency checks.
- Development of AI-supported coding assistance functions (e.g., suggestion models, highlighting of relevant text passages, active learning strategies) to accelerate coding while ensuring high quality.
- Systematic evaluation of labeling accuracy compared to traditional qualitative methods: gold standard design, inter-rater agreement, error analyses, robustness and bias checks across different text types.
- Analysis and optimization of the trade-off between accuracy and speed: measurement of throughput (labels/hour), uncertainty/confidence thresholds, human-in-the-loop strategies, cost/benefit considerations for different data sources and coding tasks.
- Establishment of a reproducible infrastructure for data preparation, training/fine-tuning, experiment tracking and documentation (including support from student assistants for implementation, testing and documentation).
- Independent completion of a dissertation on AI-supported text coding and active participation in publications, workshops, conferences and the CRC's research and study program.

### Requirements

- Very good master's degree in computer science, data science, mathematics or a related discipline; relevant research experience (e.g., master's thesis) in the field of NLP/ML/text mining.
- In-depth knowledge of the design, analysis and implementation of algorithms for large text corpora, including efficient data pipelines and clean experimental design.
- Strong NLP skills for semantic text analysis and (semi-)automatic labeling, e.g., text/segment classification, sequence labeling, span-based labeling, information extraction (NER/relations, if relevant for coding), transformer/embedding-based models, prompting vs. fine-tuning.
- Practical experience with fine-tuning and evaluation of modern language models for annotation (e.g., Adapter/LoRA, weak supervision, distillation), including reproducible setups (seeds, splits, ablations).
- Experience with annotation & coding workflows (ideally INCEpTION) and quality assurance: guideline development, consistency checks, conflict resolution, metrics such as precision/recall/F1, calibration, agreement (e.g., κ/α), systematic error analysis (label mix-ups, domain artifacts).
- Ability to develop human-in-the-loop strategies that balance accuracy and speed: active learning (uncertain cases first), smart sampling, confidence thresholds, gradations (auto-label/review/manual), measurement and decision logic for throughput vs. quality.
- Proficiency in programming (especially Python) and use of ML/NLP tooling (e.g., PyTorch/Transformers, spaCy), plus fundamentals in MLOps/experiment tracking (e.g., DVC/MLflow/Git).
- Very good written and spoken German and English skills.

We expect a high degree of initiative and intrinsic motivation, as well as a keen interest in creative problem solving using the latest technologies in the field of education. Candidates should demonstrate personal commitment, teamwork skills and the ability to work independently in an interdisciplinary environment.

We offer a stimulating and rewarding work environment and a broad network in research and practice, as well as access to Goethe University's GRADE continuing education program. We also offer you a diverse and creative role in an interdisciplinary team with a pleasant working atmosphere and plenty of opportunities to contribute and implement your own innovative ideas. This role offers you the opportunity to focus on the practical implementation and research of new technologies and possibilities for digitally supported teaching and learning in large and small projects.

Flexible working hours and a daycare center at the DIPF provide excellent conditions for balancing work and family life. Employees have the option of purchasing a discounted Jobticket Germany.

DIPF is certified with the audit berufundfamilie+vielfalt seal, promotes equality for all employees and welcomes applications regardless of ethnic, cultural or social origin, age, religion, ideology, disability, gender and sexual identity. Severely disabled people are given special consideration if they are equally suitable. A reduction in working hours is generally possible, taking into account the interests of the company.

For more information about the position, please contact Prof. Dr. Hendrik Drachsler at e-mail address h.drachsler@dipf.de. Please send your written application with the usual documents in electronic form and summarized in a pdf document, quoting **reference no**. **IZB 5115-26-01 by 09.02.2026** to:

Prof. Dr. Hendrik Drachsler
[bewerbung-tba@dipf.de](https://mailto:bewerbung-tba@dipf.de)
DIPF | Leibniz Institut for Research and Information in Education
Rostocker Straße 6, 60323 Frankfurt am Main

### Datenschutz-Hinweis

[Erstinformationen zum Datenschutz über die Verarbeitung Ihrer Daten nach Art. 13 DS-GVO](https://www.dipf.de/de/dipf-aktuell/pdf-aktuelles/datenschutz-erstinformation-de.pdf) (pdf).