2 months ago
Remote, United Kingdom +3 moreSenior
Responsibilities
- Design, develop, and maintain scalable edge infrastructure for deploying models to global device fleets.
- Own the model compilation pipeline that produces optimized, hardware-specific inference engines.
- Manage TensorRT compilation, FP16/INT8 precision trade-offs, calibration, and engine validation.
- Build automated production-device testing and telemetry pipelines for model drift, thermal impact, and inference latency.
- Develop resilient update mechanisms for low-bandwidth environments, limited storage, and network failures.
- Collaborate with data scientists, embedded engineers, and product managers on hardware integrations.
- Mentor teammates and establish best practices for Python tooling, infrastructure-as-code, and CI/CD.
Requirements
- Production experience building and operating model-deployment pipelines with Docker and Linux systems.
- Hands-on experience compiling and optimizing machine-learning models for embedded hardware, ideally with TensorRT.
- Understanding of precision, quantization, calibration, and inference-engine validation at scale.
- Ability to design resilient architectures for heterogeneous device fleets, including canary releases and safe rollbacks.
- Ability to collaborate with researchers and low-level embedded engineers to translate constraints into solutions.
- Experience with the NVIDIA edge ecosystem, including Jetson Orin, DeepStream SDK, and TensorRT, is preferred.
- Familiarity with video pipelines, GStreamer, or ffmpeg is preferred.
- Experience with AWS IoT Greengrass, Balena, or custom over-the-air and fleet-management solutions is preferred.
- Interest in sports technology, video analytics, or performance metrics is preferred.
Benefits
- Flexible vacation time, company-wide holidays, meeting-free days, and remote work options.
- Professional development resources and career growth opportunities.
- Medical and retirement benefits depending on location.
- Employee Assistance Program and employee resource groups supporting wellbeing.
- Role may be based near the London or Barcelona offices, with remote options in the UK and Spain.
Categories
About Hudl
Hudl builds sports video and analytics tools for coaches, athletes, and administrators, combining SaaS software with smart cameras, instant replay, recruiting, and livestreaming features. The company sells subscriptions and hardware such as Hudl Focus to teams from youth to professional levels to capture, analyze, and share performance. Founded in 2006 and headquartered in Lincoln, Nebraska, Hudl is privately held and its platform is used by over 230,000 teams worldwide.