Machine Learning Engineer, Synthetic Data

Overview

You will advance GAIA-class world models and synthetic-data pipelines to accelerate autonomous driving development. You’ll bridge research and engineering to scale generation, post-train models, and integrate synthetic data into the training stack. The role sits at the intersection of ML research and platform engineering, shaping capabilities that directly impact real-world driving models. You’ll work with cross-functional teams to improve training speed, evaluation, and deployment of driving policies. This is a unique opportunity to influence end-to-end autonomy through scalable, data-centric AI.

Pay / Benefits

• Salaries benchmarked against market
• Equity/ownership opportunities
• Relocation support and visa sponsorship
• Hybrid working with core hours
• Learning and development budgets
• Comprehensive health and wellbeing benefits

Responsibilities

• Post-train or ablate GAIA rig-transfer or pose-transfer checkpoints and diagnose failures
• Launch and manage large-scale synthetic data generation jobs (tens of thousands of segments); optimize GPU capacity and generation parameters
• Integrate synthetic data into training pipelines (binzarise, sampling hooks, RL/read suite, on-road diffs)
• Expand capabilities to new vehicle embodiments and safety-critical scenarios; collaborate with OEM teams
• Collaborate with world-model researchers, infra, and driving-model owners to ensure generation, evaluation, and training form a cohesive system
• Engage in paper-like research alignment; contribute to outcomes that affect checkpoints or data mixes shipped

Key requirements

• 4+ years in applied ML / research engineering with training and shipping neural nets
• Strong Python and PyTorch; comfortable with GPU training, debugging, and reading model code
• Hands-on experience with video, generative, or world models (diffusion/flow-matching/autoregressive video)
• Working knowledge of cameras and 3D geometry (multi-camera rigs, intrinsics/extrinsics, warps/reprojection)
• Experience translating generated or simulated data into downstream models and measuring impact (mixes, ablations, failure analysis)
• Ability to operate generation/training at real scale (multi-GPU, workflow orchestration) and make processes reliable
• Collaborative, experimental working style with researchers and platform engineers; ownership mindset

• collaborative
• problem-solving
• ownership
• PyTorch
• GPU/large-scale training
• video generation or world models (diffusion/flow-matching/novel-view synthesis)

…