ECCV, Malmö: workshop presentations on procedural understanding, robot learning, and egocentric AI, including oral presentations at ACVR and X-Reason.
Background & experience
About me.
Connecting machine learning research with the demands of real-world systems.
I am a PhD candidate at TU Berlin and a Research Associate at Fraunhofer IPK. My research connects multimodal perception, procedural understanding, and embodied action: how can a system learn from human demonstrations, maintain an understanding of a task, and act reliably as conditions change?
Before moving into machine learning research, I worked on vehicle simulation and engineering with Tata Technologies and Jaguar Land Rover in India and the UK. That experience continues to shape my interest in rigorous evaluation and practical deployment.
Research & mentorship
At Fraunhofer IPK, I develop computer vision and multimodal learning systems, collaborate across research and industrial projects, and supervise student research. My mentorship includes five Bachelor’s/Master’s theses and nine research internships.
I also contribute to the AI Grid community in Berlin and have served as a reviewer or sub-reviewer for CVPR and ECCV 2026.
Sharing research
Talks & activities
TU Berlin seminar: Vision-Language-Action Models for Industrial Robotic Applications.
ML AfterMath Workshop, ECDF Berlin: IndEgo: Industrial Scenarios and Collaborative Work for Egocentric Assistants.
NeurIPS: presentation of IndEgo in the Datasets & Benchmarks track.
Invited talk at TU Berlin: egocentric and exocentric vision for industrial applications.
AI Grid Summit, ECDF Berlin: continual learning in real-world scenarios.
Winner, computer vision hackathon on food waste detection at Hasso Plattner Institute, sponsored by Voxel51.
Sorbonne University, Paris: developing industrial egocentric assistants through continual learning and multimodal understanding.
Heise KI-Update Deep Dive: interview on AI glasses in industry. Listen ↗
Berlin Science Week research talk and demonstration; AI Grid Hackathon winner; Science Slam Hamburg winner; technology demonstrations at Control, Stuttgart.
Fraunhofer Vision invited talk on energy-aware continual learning, and six internal VisionRead research talks.