After careful consideration, we decided to select the following 18 proposals for finalist presentation. Congratulations to the students that qualified!
| Student Name | University | Innovation Title | Supervisors |
| Amin Abdulrahman | Max Planck Institute for Security and Privacy | Side-Channel-Secure Post-Quantum Crypto on an Open-Source Silicon Root of Trust | Peter Schwabe |
| Zhuo Chen | University of Birmingham | Touch from Appearance: Learning Comprehensive Object Physics for Human-Like Robotic Manipulation | Hyung Jin Chang Aleš Leonardis |
| Yupeng Chen | University of Oxford | Establishing Safe Protocols for Agent-X Interactions | Adel Bibi Philip Torr |
| Martin Fink | Technical University of Munich | Provable Security for JIT Compilers | Pramod Bhatotia |
| Hanna Foerster | University of Cambridge | Toward Secure-By-Default Computer Use Agents | Robert Mullins |
| Mar Gonzàlez I Català | University of Cambridge | A Geometric Theory of Autoregressive Reasoning | Pietro Liò |
| Jiajun He | University of Cambridge | Robust Diffusion Model Control for Multiple-Constraint Inverse Problems with Replica Exchange | José Miguel Hernández-Lobato |
| Tom Jacobs | CISPA | Controlling Implicit Regularization for Sparse Transformers | Rebekka Burkholz |
| Tobias Kovats | ETH Zürich | Finding Microarchitectural Leakage in Concurrent Systems Before Tapeout | Kaveh Razavi |
| Zongqian Li | University of Cambridge | Scaling Agent Skills and Cowork Environments: Training Generalized On-Device Agents via Reinforcement Learning | Nigel Collier |
| Diganta Misra | Max Planck Institute for Intelligent Systems | Hodge-Regularized GRASP for Noisy Preference Post-Training | Antonio Orvieto |
| Abhinandan Pal | University of Birmingham | Neural Model Checking for Hardware Verification | Mirco Giacobbe |
| Jeroen Robben | KU Leuven | Towards Practical Stateful Fuzzing of Network Protocol Implementations | Mathy Vanhoef |
| Naila Sebastián Esandi | INRIA | A Theoretical Framework for Zero-Shot Reinforcement Learning | Vianney Perchet |
| Wenfang Sun | University of Amsterdam | Trustworthy Video Agents on a Budget: Multi-Round Reasoning with Physics-Grounded World Models | Cees G.M. Snoek |
| Christopher Wewer | Max Planck Institute for Informatics | Towards Scene State Tokenization for World Models | Bernt Schiele Jan Eric Lenssen |
| Xianghui Xie | University of Tübingen | From Videos to Policies: Autonomous Robot Learning from Human Demonstrations | Gerard Pons-Moll |
| Jiaxu Xing | ETH Zürich | Learning Self-Improving Robotic Control Policies in the Real World | Davide Scaramuzza |
