Research engineer Perspective-Grounded Spatial Reasoning in Multimodal Large Language Models
Inria · Paris
Research instituteonsite
At a glance
Highlights
- Build a diagnostic benchmark for perspective-taking failures in video MLLMs.
- Develop 3D-grounded spatial reasoning aligned with pretrained 3D models.
- Design novel solutions for allocentric viewpoint reasoning from video.
Ideal candidate
A curious researcher passionate about multimodal models and 3D geometry who enjoys diagnosing hard failures and turning insights into principled, publishable solutions.
Requirements
- Programming skills
- English fluency
- Good communication skills
Lab profile
About Inria
French national research institute advancing digital sciences, computing, AI and applied mathematics.
What they build
- Digital sciences and computing research
- Artificial intelligence for microscopy and video analysis
- Quantum information theory
Known for
- Joint teams with Paris-Saclay University and Institut Polytechnique de Paris
- Embedded in Paris-Saclay science and technology cluster
- Rich R&D ecosystem with startups, industry and clusters
- International collaborations and multinational scientific staff
Key facts
- Founded in 1967
- Headquartered in France
- Nine research centres across France
Posted · Last seen