Research Associate in Safe Reinforcement Learning

Overview

As a Postdoctoral Research Associate in the FMAI lab, you will lead innovative research at the intersection of formal methods and safe reinforcement learning. You will develop certified RL approaches and verification via proof certificates for cyber-physical systems, contributing to the MASA-Safe-RL library. You will collaborate with a cross-disciplinary team to publish in top venues and advance safe, scalable AI for safety-critical applications. This role offers a fully funded opportunity to shape methods that bridge learning and formal guarantees.

Pay / Benefits

• fully funded position
• sector-leading salary and package
• 41 days off per year
• generous pension schemes
• career support for researchers
• diverse and collaborative culture

Responsibilities

• Conduct original research in Formal Methods for Safe RL and its cyber-physical applications
• Develop novel algorithms leveraging proof certificates
• Collaborate with RL, formal methods, strategy synthesis, and multi-agent systems researchers
• Contribute to the MASA-Safe-RL library and related open-source efforts
• Publish results in top-tier conferences and journals
• Mentor or supervise junior researchers where appropriate
• Disseminate findings through talks and internal seminars

Key requirements

• PhD in Computer Science or Mathematics
• strong track record in top conferences/journals in Formal Methods or Reinforcement Learning
• excellent skills in mathematics and foundations of deep learning
• experience with deep learning libraries (PyTorch or JAX)
• ability to design and evaluate formal verification and learning-based methods

• self-driven
• strong collaboration
• effective communication
• Formal methods
• Reinforcement learning
• Proof certificates

…