Throughout the program, participants engaged with a wide range of topics, including:

• Automatic Differentiation in Optimization: Guillaume Dalle’s tutorial highlighted the role of forward and reverse AD in embedding solvers within neural architectures for decision-focused learning.
• Combinatorial Optimization–Augmented Models: Sessions introduced Fenchel-Young losses and exponential-family formulations, applied to supply chain and network design problems.
• Learning for Nonlinear Solvers: Bissan Ghaddar shared breakthroughs on ML-guided branching and cutting for mixed-integer nonlinear programs, achieving speedups on MINLPLib and QPLIB.
• Reinforcement Learning for Stochastic Decisions: Wouter van Heeswijk linked dynamic programming and policy-gradient RL (TRPO, PPO), with applications in logistics and finance.
• Interactive Coding Workshops: Léo Baty guided participants in Julia and Python sessions to implement solver-integrated learning pipelines.

Evening social events—including traditional Swabian dinners at Lehners Wirtshaus and lakeside gatherings at Insel-Hotel—sparked rich discussions on decision-centric AI, risk-aware modeling, and multi-agent learning.

We thank Prof. Maximilian Schiffer and Prof. Axel Parmentier for organizing an exceptional and future-oriented program. The insights and connections formed in Heilbronn will continue to shape our research moving forward.