VLA Pre-training Engineer (Deep Learning)

📋 Description

• Train and refine policies via behaviour cloning and RL across data-to-deployment.
• Collaborate with Data Collection to define good data, identify failure modes, ensure coverage.
• Coordinate with external partners to secure high-quality pretraining data.
• Run pre-/mid-/post-training on the VLA stack; explore modalities and architectures.
• Build and maintain pipelines: ingest synthetic data and teleop logs, label data, curate datasets, surface failures for retraining.
• Work with MLOps & Data Platform teams to scale distributed training and edge inference.

🎯 Requirements

• 3+ years building deep-learning systems with shipped models.
• Hands-on with LLMs, VLMs, or image/video generative models.
• DL infra experience: streaming data, checkpoints, state mgmt, distributed training.
• Strong Python + PyTorch/JAX; profile, debug numerics, maintainable code.
• Familiarity with modern software engineering practices.
• Document experiments clearly and communicate trade-offs crisply.

🎁 Benefits

• Competitive equity: stock options with meaningful upside.
• 30+ paid days off, including annual leave and UK holidays.
• Private healthcare, including virtual and in-person care.
• Pension scheme with 8% total contribution (5% employee, 3% employer).
• Free daily breakfast, catered lunch, and snacks in-office.
• Real ownership: direct access to leadership and product direction.