> Markdown version of [/jobs/ext/1563126-job-offer](https://www.wearedevelopers.com/jobs/ext/1563126-job-offer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Job offer - **Company:** Eindhoven University of Technology - **Location:** Netherlands (Remote available) - **Salary:** €4,241.0 - **Contract:** Temporary contract - **Skills:** Artificial Intelligence, Collaborative Learning, Data Governance, Github, Python (Programming Language), Machine Learning, Natural Language Processing, Tensorflow, Data Streaming, Jupyter Notebook, High Performance Computing, Pytorch, Large Language Models, Prompt Engineering, Generative AI, Information Technology, HuggingFace, Free and Open-Source Software, Multiaccess Edge Computing, Virtual Agents - **Published:** July 11, 2026 - **Apply:** https://www.academictransfer.com/en/jobs/362497/postdoc-in-ai-driven-adaptive-learning-systems-clara-project/apply/ ## About the Role We are looking for a technically strong and intellectually curious researcher. A rigorous computational background is essential; experience with educational contexts is valuable but secondary to technical excellence., * A PhD degree in Computer Science, Artificial Intelligence, or a closely related technical field. * Demonstrated expertise in large language models (LLMs), natural language processing (NLP), and/or machine learning, with a verifiable track record (e.g., publications, thesis, or open-source contributions). * Proficiency in Python and relevant ML frameworks (e.g., PyTorch, HuggingFace Transformers, LangChain). * Experience with model fine-tuning, prompt engineering, retrieval-augmented generation (RAG), and with agentic AI and AI agents. * Familiarity with responsible AI principles, including fairness, transparency, and data governance. * Proven experience with supercomputing / HPC environments. * Strong academic writing and communication skills in English. * Ability to work effectively in an interdisciplinary team, including colleagues from social and educational sciences. Desirable qualifications The following qualities are not required but will significantly strengthen an application: * Experience working with or conducting research in educational settings - hands-on knowledge of how learning environments operate greatly facilitates collaboration with the pedagogical team and the PhD candidate. * Familiarity with multimodal data (audio, video, interaction logs) and time-series analysis of social interaction. * Experience with or interest in agentic AI systems and real-time inference pipelines. Awareness of Challenge-Based Learning or comparable active learning frameworks. ## Description Eindhoven University of Technology invites applications for a postdoctoral research position within the recently NRO-funded project AI as a Social Agent to Support Group Learning Processes (CLARA). This interdisciplinary project investigates how AI - specifically large language model (LLM)-based agents - can act as adaptive social agents to support students' collaborative learning in Challenge-Based Learning (CBL) environments. You will be embedded in the Department of Industrial Engineering & Innovation Sciences (IE&IS) and form the technical core of the CLARA consortium, which brings together researchers from TU/e, the University of Twente, and Maastricht University. Working closely with a PhD candidate and the project's supervisory team, you will design, train, and iteratively refine the CLARA AI agent - bridging cutting-edge machine learning methods with empirical insights from the educational arm of the project. A central technical challenge guides this position: How can an LLM-based AI social agent be designed, fine-tuned, and deployed to detect socio-cognitive and socio-emotional triggers in student group work, and deliver contextually appropriate scaffolding in real time? Research tasks and key deliverables Rather than following a fixed phase sequence, you are expected to make substantive contributions across the following six areas throughout the appointment: * Trigger detection system. Design and implement an NLP/LLM-based system capable of identifying socio-cognitive and socio-emotional triggers in student group interaction data (text, audio, and multimodal streams), drawing on the HASRL framework and the empirical taxonomy developed by the PhD candidate. * Model training and fine-tuning. Fine-tune large language models on annotated educational datasets collected during the project, ensuring the agent's responses are pedagogically valid, contextually appropriate, and consistent with collaborative learning theory. * Scaffolding mechanism design. In close collaboration with the PhD candidate and educational supervisors, develop and evaluate adaptive scaffolding strategies that the AI agent delivers as interventions, refining them iteratively based on classroom data and pedagogical feedback. * Classroom implementation and evaluation. Support and co-lead pilot studies in real CBL classrooms; contribute to data collection, analysis, and interpretation of the agent's performance, attending to both technical metrics and educational outcomes. * Responsible AI and fairness auditing. Conduct algorithmic fairness validation of the CLARA system, develop documentation on data governance and GDPR compliance, and contribute to the project's open-science outputs, including containerized model workflows. * Dissemination and scientific output. Publish findings in peer-reviewed journals, present at leading conferences, and contribute to practice-oriented outputs and knowledge transfer activities within the NRO consortium., * One representative publication or code portfolio demonstrating relevant technical work. * Research and technical challenge response (see below). Technical challenge (required) As part of the application, candidates are asked to complete a short technical challenge designed to give a concrete and objective view of their skills and thinking. It has two parts: Part A - Implementation (approx. 1-3 hours). Suppose you have access to a data stream from an upstream pipeline that classifies students' socio-emotional and socio-cognitive states (e.g., confusion, frustration, disengagement, neutral) based on their speech. Write a short Python script or notebook that (1) processes a sample of this data (you may use a small synthetic or publicly available dataset), (2) applies a rule-based or model-based method to identify a moment where an AI agent's support intervention would be warranted (a 'trigger event'), and (3) generates or selects an appropriate scaffolding message for the group. Include brief comments explaining your design choices. Part B - Reflection (max. 400 words). Briefly discuss: What kinds of support would be appropriate for a group in this situation, and what would be inappropriate or potentially harmful? What are the key limitations of using speech emotion data for this purpose, and how would you address them in a responsible deployment? There is no single correct answer. We are looking for clear reasoning, technical competence, and awareness of the educational and ethical dimensions of the task. Submissions may be in the form of a GitHub link, Jupyter notebook, or a PDF. ## Related Videos - [Innovating Developer Tools with AI: Insights from GitHub Next](https://www.wearedevelopers.com/videos/1268-innovating-developer-tools-with-ai-insights-from-github-next) - [AI in the Open and in Browsers - Tarek Ziadé](https://www.wearedevelopers.com/videos/1787-ai-in-the-open-and-in-browsers-tarek-ziade) - [Photonic Computing: Programming a New Class of AI Accelerators (incl. Live Coding)](https://www.wearedevelopers.com/videos/100196-photonic-computing-programming-a-new-class-of-ai-accelerators-incl-live-coding) - [Machine learning in the browser with TensorFlowjs](https://www.wearedevelopers.com/videos/155-machine-learning-in-the-browser-with-tensorflowjs) - [Bringing AI Model Testing and Prompt Management to Your Codebase with GitHub Models](https://www.wearedevelopers.com/videos/1536-bringing-ai-model-testing-and-prompt-management-to-your-codebase-with-github-models) - [Serverless deployment of (large) NLP models ](https://www.wearedevelopers.com/videos/158-serverless-deployment-of-large-nlp-models) ## Related Articles - [How to Become an AI Engineer](https://www.wearedevelopers.com/magazine/331-how-to-become-an-ai-engineer) - [What Are Large Language Models?](https://www.wearedevelopers.com/magazine/304-what-are-large-language-models) - [Got AI ideas but no money? Here are 10 free ways to level up your AI skills with Google Cloud](https://www.wearedevelopers.com/magazine/600-got-ai-ideas-but-no-money-here-are-10-free-ways-to-level-up-your-ai-skills-with-google-cloud) - [Dev Digest 121 - AI goes offline](https://www.wearedevelopers.com/magazine/456-dev-digest-121-ai-goes-offline) - [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development) - [Prompt Engineering is a Job of the Past](https://www.wearedevelopers.com/magazine/342-prompt-engineering-is-a-job-of-the-past)