Master's student at TU Munich and research assistant in Prof. Daniel Cremers' Computer Vision & AI group. Work focuses on inference-time
alignment of diffusion and flow models — specifically reward-agnostic methods that require no fine-tuning or differentiable rewards. Author of
a workshop paper at ICLR 2026.
Education
Technical University of Munich (TUM), Germany
- Current GPA: 1.4 (German scale)
- Focus: Machine learning, generative models, inference-time optimization
- Exchange: Chalmers University of Technology, Gothenburg, Sweden (Sep 2024 – Jan 2025)
Technical University of Munich (TUM), Germany
- Final GPA: 2.5 (German scale)
- Specialization: Artificial Intelligence and Machine Learning
Experience
Chair for Computer Vision & Artificial Intelligence (CVAI), TU Munich
Research on inference-time alignment of diffusion and flow models in Prof. Daniel Cremers' group. Contributed to the ICLR 2026 workshop
paper on trust-region noise optimization. Concurrent with Master's thesis in the same group.
Siemens AG
Developed a 3D feature matching system using CNN embeddings to automate part retrieval in manufacturing databases, bridging deep learning
with industrial applications.
Publications
Schweiger, N., Ram, K., & Cremers, D. (2026).
Trust-Region Noise Search for Black-Box Alignment of Diffusion and Flow Models.
ReALM-GEN Workshop @ ICLR 2026, Rio de Janeiro, Brazil.
Awards & Achievements
Built Caire, an app designed to overcome language barriers between nurses and patients using AI to extract medical information
from natural speech. Awarded a €2,000 prize.
Skills & Information
- Programming
- Python (Advanced), PyTorch (Advanced), C/C++ (Intermediate), Matlab (Intermediate), LaTeX
- Languages
- German (Native), English (C1 – Professional working proficiency)
- Interests
- Football, Strength training & running, Guitar (10+ years), AI & Natural sciences
