Senior MLOps Engineer - Edge

The Senior MLOps Engineer will join the Hardware Group to build and scale the machine learning infrastructure that powers Focus, Hudl's line of smart cameras. The role owns the edge deployment pipelines that transport neural networks from training clusters to tens of thousands of devices globally and contributes to the platform that compiles trained models into optimised inference engines for devices such as the Jetson Orin. The engineer will help build the 'nervous system' for the next generation of automated sports capture and work with Data Scientists, Embedded Engineers, and Product Managers.

• Design, develop, and maintain scalable edge infrastructure and delivery systems that enable models to be deployed to fleets of devices.
• Build and maintain the model compilation platform that takes trained models and produces optimised, hardware-specific inference engines, managing TensorRT compilation, precision trade-offs (FP16/INT8), calibration
• And engine validation to ensure models run reliably and efficiently on target devices.
• Collaborate with Data Scientists, Embedded Engineers, and Product Managers to ensure smooth integration of complex features and capabilities.
• Implement infrastructure to silently test candidate models on production devices and build telemetry pipelines to monitor drift, thermal impact, and inference latency in the wild.
• Tackle the unique constraints of the edge by building resilient update mechanisms for low-bandwidth environments, optimising for limited storage, and ensuring devices recover gracefully from network failures.
• Share expertise to establish best practices in Python tooling, Infrastructure-as-Code, and CI/CD, guiding the team toward a more robust, automated future.

• Production MLOps expertise: a key role in building and operating pipelines that deploy models to production, with deep experience in CI/CD, containerisation (Docker), and Linux systems.
• Collaborative approach: understands that shipping to hardware is a team sport and can communicate effectively with researchers and low-level embedded engineers to translate constraints into solutions.
• Systems thinking: can design architectures that handle failure gracefully and understands the implications of deploying to 10,000 heterogeneous devices, including managing risk via canary releases and safe rollbacks.
• Bias towards action: sees the role as solving problems, filling gaps, and taking initiative as needed to help the team win together.

• Edge inference and compilation know-how: hands-on experience compiling and optimising models for embedded hardware, ideally with TensorRT, and understanding the practical implications of precision, quantisation, and engine validation at scale.
• Experience with the NVIDIA edge ecosystem (Jetson Orin, DeepStream SDK, TensorRT).
• Familiarity with video pipelines, GStreamer, or ffmpeg.
• Experience with tools such as AWS IoT Greengrass, Balena, or custom OTA/fleet management solutions.
• Interest in sports technology, video analytics, or performance metrics; the domain can be taught if not.

• Hudl would like to hire someone who lives near its offices in London or Barcelona, but is also open to remote candidates in the UK and Spain.

• Flexible vacation time above any required statutory leave.
• Company-wide holidays.
• Remote work options.
• Paid sabbatical every five years to unplug and recharge.
• Professional development resources and opportunities to keep growing.
• Offices designed with employees in mind, plus the tech needed to do your best work whether in the office or remote.
• Depending on location, medical and retirement benefits.
• Employee Assistance Program and employee resource groups to support mental health.

Hudl describes itself as the team behind the team, supporting the lifelong impact sports can have through lessons in teamwork and dedication, inspiring coaches, and opportunities to reach new heights.