Who are we?

Founded in 2014, Oxa is a global leader in autonomous vehicle (AV) technology, dedicated to accelerating Industrial Mobile Autonomy (IMA).

We develop advanced physical AI and robotics technology, anchored around our configurable and explainable self-driving software, Oxa Driver; development toolchain, Oxa Foundry; and fleet management software, Oxa Hub. We utilise hardware blueprints known as Reference Autonomy Designs (RADs) to enable the integration of sensors, compute and drive-by-wire systems into existing vehicles produced by OEMs.

Our solutions automate repetitive industrial driving tasks, such as the towing and carrying of goods in locations like ports, airports and manufacturing facilities, or asset and perimeter monitoring in environments such as solar farms or industrial plants. We’re helping global businesses to address critical challenges like labour shortages and rising operational costs - driving efficiency, productivity, and safety. Based in Oxford, and with offices in Canada, our engineering team is drawn from the world’s top physical AI specialists and led by originators of the field.

Your Role:

You will join a growing team of computer science and robotics experts bridging the gap between cutting-edge machine learning and production autonomy. Your work will focus on integrating ML-based reasoning models into the broader Oxa Driver™ planning and driving stack, whilst developing the simulation and metrics tools necessary for closed-loop evaluation and validation at scale.

Key Responsibilities:

• Integrate state-of-the-art machine learning (ML) based motion planning models (e.g., Behaviour Cloning, Reinforcement Learning) into the core planning and driving software stack, ensuring seamless interoperability and real-time performance.
• Develop and maintain driving simulation and scenario generation tools to stress-test planning behaviors against diverse, safety-critical edge cases.
• Design and execute closed-loop evaluation frameworks that quantify system-level performance and provide rapid feedback for model improvement.
• Bridge the sim-to-real gap between offline model training, virtual testing, and on-vehicle performance by instrumenting, monitoring, and analyzing model behavior within the simulation environment.
• Collaborate across teams to ensure that ML planning models respect the constraints and requirements of the full autonomy stack, including perception, mapping, and vehicle control.
• You will be encouraged to share your ideas with the team and the wider business.
• You will interact with other teams to learn about the autonomy system and gain exposure to all aspects of the business.