Role Overview
The internship focuses on proof‑carrying neural network controllers, integrating Vehicle, Rocq, and Marabou to build a verification pipeline that creates machine‑checkable proofs for safety‑critical neural controllers. The goal is to enable continuous updates of neural models while retaining independently verifiable correctness arguments for autonomous systems.
Responsibilities
- Develop software that integrates Vehicle, Rocq, and Marabou into a verification pipeline for proof‑carrying neural network controllers.
- Support scientists with software development for user‑ready prototypes, data analysis, and user interfaces.
- Present demos to visitors and internal stakeholders.
- Acquire and apply software development skills in a supportive, continuous‑learning environment.
Requirements
- Enrolled in a Bachelors, Masters, or PhD program in Computer Science, AI, interactive theorem provers, neural networks, PINNs, robotics, cyber‑physical systems, neuro‑symbolic systems, or a related discipline.
- Interest in formal methods, verification, and safe operation of controllers (preferred).