Sameer Ambekar is a Ph.D. researcher at the Technical University of Munich (TUM) and Helmholtz Munich, working in deep learning under Prof. Dr. Julia Schnabel. His doctoral research focuses on adapting vision-language models using Group Refinement Policy Optimization (GRPO) and test-time adaptation. He is fully funded by the Konrad Zuse School of Excellence in Reliable AI (relAI) and is an affiliate member of the Munich Center for Machine Learning (MCML). At TUM, he also supervises bachelor’s and master’s theses, as well as student projects.

Prior to his Ph.D., he received his M.Sc. in Artificial Intelligence from the University of Amsterdam (UvA), Netherlands. During this time, he worked as a research intern at the AIM Lab, UvA. His master’s thesis (48 ECTS) focused on test-time adaptation for domain generalization by generating models and labels through variational meta-learning, with an emphasis on meta-learning, variational inference for efficient multi-source training, and transferable feature learning.

Before his MSc AI, he worked as a Research Assistant at IIT Delhi (IITD), where he addressed unsupervised domain adaptation using methods such as variational generative latent search in the VAE latent space. His broader research interests lie in unsupervised learning, particularly through meta-learning and variational inference, with a focus on learning efficient and transferable representations. He conducts the seminar “From General to Clinical: Adapting Foundation Models for Medical Images” for master’s students at the Technical University of Munich in both the Winter Semester 2025/26 and the Summer Semester 2026.

• Open Theses and Projects: I am particularly interested in topics such as Adapting foundation models, Vision and Language Models (VLMs), Fine-tuning, Reasoning, and related areas (not limited to these). For MSc thesis or project supervision, you are welcome to reach out to me directly via email or website - https://ambekarsameer.com.