MDSI & relAI Pitchtalks at SAP Labs Munich

On Tuesday, September 22 from 16:00-17:30 (CET), we are delighted to host pitchtalks with MDSI & relAI at the SAP Labs Munich Campus Auditorium (AE.76).

What?  Join us for a lively meetup with leading researchers and SAP experts to connect, discuss, and explore new ideas
When? September 22, 2026 - Doors open at 15:30, talks start at 16:00 
Where? SAP Labs Munich Campus, Friedrich-Ludwig-Bauer-Straße 5, 85748 Garching, Germany
Questions? iuc-tum@sap.com

Register here and join us for the pitchtalks with MDSI & relAI. Confirmed topics include: 

Bhavatarini Kumaravel (TUM): A Graph-Based Deep Q-Learning Agent for Grammar-Guided Structural Form-Finding
MDSI Core Member, TUM Chair for Structural Design, TUM School of Engineering and Design
The work investigates how graph-based deep reinforcement learning can be combined with a structural grammar in finding novel and materially efficient structural design solutions.

Annika Schneider (TUM): Decision-aligned Evaluation of Uncertainty Quantification
relAI Fellow, TUM Associate Professorship of AI for Scientific Modelling
This work investigates the question of how probabilistic models should be evaluated to ensure they perform in downstream decisions.

Sameer Ambekar (TUM): Adressing Distribution Shifts: The Shift from Static to Thinking-based Vision-Language Models
relAI Fellow and MDSI Core Member, TUM Chair for Computational Imaging and AI in Medicine
Distribution shifts have remained a persistent challenge over the years, with solutions spanning from domain adaptation to test-time training and, most recently, post-training mechanisms for vision-language models. This is because models inevitably encounter unseen data at inference time, data they were never trained for.

This talk at SAP by Sameer will focus on addressing this problem, tracing it from convolutional architectures to modern vision-language models and reasoning models, and connecting these stages through a consistent pattern: models that allocate additional computation at test time, whether through adaptation or reasoning, generalize more reliably than those relying solely on fixed, pre-trained parameters. Sketching through these topics from his PhD research, the talk will provide an overview of addressing unseen data at test time through training, reasoning, and post-training mechanisms.

_____

On MDSI: The digital revolution is transforming societies, economies, and even science and its paths to knowledge. The Munich Data Science Institute (MDSI) anticipates, accompanies and shapes this change. The MDSI is TUM’s central interface and innovation hub for questions and solutions arising from Data Science, Machine Learning and Artificial Intelligence. MDSI is connecting people and ideas across disciplines.

On relAI: The Konrad Zuse School of Excellence in Reliable AI (relAI) envisions training future AI experts who uniquely blend technical proficiency with a deep understanding of AI reliability. RelAI's groundbreaking and innovative AI program aims to educate top international candidates in the comprehensive development of reliable AI systems. This includes scientific knowledge, business acumen, and industrial experience, preparing them for roles in both industry and academia. Additionally, relAI conducts cutting-edge research to ready AI for deployment in critical applications.