Summary
Vrije Universiteit Amsterdam (VU Amsterdam) invites applications for a fully-funded, 2-year Postdoctoral Researcher position focused on advancing neurosymbolic reasoning and commonsense abstraction within multimodal world models. Funded by the NWO AiNed talent project “Human-centric AI with common sense”, the successful candidate will work under the co-supervision of Dr. Filip Ilievski, Dr. Emile van Krieken (VU Amsterdam), and Prof. Cees Snoek (University of Amsterdam / UvA) as part of the ELLIS Unit Amsterdam.
Postdoc Position: Neurosymbolic Reasoning in Multimodal World Models
Designation
Postdoctoral Researcher (Postdoc) / Researcher 4
Key Information Table
| Attribute | Details |
| Institution | Vrije Universiteit Amsterdam (VU Amsterdam) |
| Faculty / Department | Faculty of Science / Department of Computer Science (Learning and Reasoning Group) |
| Project Title | Human-centric AI with common sense (NWO AiNed Talent Project) |
| Position Type | Full-Time (1.0 FTE), 2-year fixed-term contract |
| Expected Start Date | Q1 2027 (preferred) |
| Gross Salary | €3,706.00 to €5,760.00 gross/month (Scale 10, UFO Researcher 4) |
| Location | Amsterdam, Netherlands (NU Building, Zuidas Campus) |
| Application Deadline | October 10, 2026 |
Research Area
- Artificial Intelligence & Machine Learning: Multimodal World Models, Neurosymbolic AI, Foundation Models.
- Core Topics: Commonsense reasoning, spatio-temporal and causal abstraction, visual grounding, explicit knowledge representation, and learning from limited human-annotated data.
Location
Vrije Universiteit Amsterdam (VU Campus)
Nieuwe Universiteitsgebouw (NU), De Boelelaan, Amsterdam (Zuidas district), North Holland, Netherlands.
Eligibility & Qualifications
Applicants should meet the following criteria:
- Degree: A completed (or near-completed) PhD in Machine Learning, Computer Vision, Natural Language Processing, Neurosymbolic AI, or a closely related field.
- Research Track Record: Demonstrable record of publishing in premier AI/ML conferences or journals (e.g., NeurIPS, CVPR, ACL, ICLR, AAAI, IJCAI).
- Technical Skills: Strong programming and implementation skills in Python and modern deep learning frameworks (e.g., PyTorch) working with multimodal foundation models.
- Competencies: Ability to work collaboratively within international teams, co-supervise PhD/MSc students, and communicate fluently in written and spoken English.
Scholarship / Fellowship Description
While current multimodal world models show rapid progress, they struggle with commonsense reasoning, visual grounding, and modeling physical and social causality.
Core Objectives & Responsibilities:
- Invent, develop, and benchmark methods for explicit abstraction and dynamic neurosymbolic world models.
- Evaluate reasoning capabilities over space, time, and causality following cognitive principles.
- Co-supervise PhD and graduate students in the Learning and Reasoning group.
- Publish and present findings at top-tier international workshops and conferences.
- Contribute moderately to academic teaching and mentorship activities (e.g., guest lectures, lab sessions).
Benefits & Compensation:
- Remuneration: Monthly gross salary between €3,706 and €5,760 depending on experience.
- Annual Bonuses: 8% holiday allowance and 8.3% end-of-year bonus.
- Leave: 232 holiday hours per year (for a 38-hour week), expandable to 328 hours with a 40-hour work schedule.
- Perks: Solid ABP pension scheme, commuter expense contributions, and relocation assistance through the VU International Staff Support office.
How to Apply
Applications must be submitted online through the official portal:
- Visit the VU Amsterdam Vacancy Page.
- Click the “Apply” button.
- Upload your Curriculum Vitae (CV) and a targeted Cover Letter / Letter of Motivation.(Note: Applications sent directly via email are not accepted.)
Contact for Queries:
- Academic Enquiries: Dr. Filip Ilievski (Assistant Professor) – f.ilievski@vu.nl
- Relocation / Visa Inquiries: International Staff Support – internationalstaffsupport.soz@vu.nl
Last Date for Apply
October 10, 2026 (23:59 CET)








