Senior Machine Learning Engineer, AI Performance

Overview

In this role you own end-to-end delivery of production-ready model releases for Wayve’s autonomous driving platform. You train and iterate PyTorch models with a hypothesis-driven approach, optimize for on-vehicle runtime constraints, and collaborate with performance teams to ensure reliable handoffs from training to deployment. You’ll tackle latency and efficiency challenges with practical techniques, shaping the transition from research to real-world deployment. This is a hands-on role in a fast-moving, ownership-driven team with significant impact on safety and performance.

Pay / Benefits

• salary benchmarks
• meaningful equity
• relocation support and visa sponsorship
• hybrid working with core hours
• training and conference budgets
• comprehensive health benefits and wellbeing support

Responsibilities

• Own end-to-end delivery of model releases, from requirements to deployment readiness
• Train and iterate PyTorch models using hypothesis-driven evaluation
• Debug performance regressions and root-cause issues
• Collaborate with adjacent ML and performance engineering teams for handoffs and bottleneck alignment
• Communicate delivery timelines, trade-offs, and readiness criteria
• Meet tight on-vehicle runtime constraints as models scale
• Apply quantisation, distillation, and low-rank techniques where sensible
• Shape handoff between training, evaluation and deployment for fast, reliable ships
• Work across abstraction levels from model behavior to runtime implications
• Contribute to multi-team efforts to optimize overall system performance

Key requirements

• Proven experience improving production systems with tight constraints (latency, memory, power/cost)
• Hands-on experience training and iterating DL models in PyTorch
• Proficiency with at least one stack or toolchain (TensorRT, CUDA, Qualcomm QNN, Triton, OpenCL) and ability to learn new ones quickly
• Comfort moving between high-level model behavior and low-level kernel/runtime execution
• Familiarity with model optimisation concepts such as quantisation and/or distillation

• strong collaboration and cross-functional communication
• ownership mindset and proactive problem solving
• ability to reason about trade-offs and communicate impact to stakeholders
• PyTorch
• TensorRT
• CUDA

…