Skip to main content
TALENT247 // SPECIALIST AI EXECUTIVE SEARCH & TECHNICAL RECRUITMENT
ROB · Robotics & Embodied AI

Senior ML Engineer, Neural World Simulation & Video Diffusion

London, UK · Full-time · Hybrid

AI Specialism

Robotics & Embodied AI

Workplace

Hybrid

Location

London, UK

Compensation

£115k – £160k

annual package

Role Dossier & Specification

REF: SENIOR-ML-ENGINEER

THE CLIENT & MANDATE

Our client is a category-defining embodied AI company in London building mapless, end-to-end foundation models for autonomous vehicles. To evaluate and train safety-critical driving policies beyond recorded road logs, their Simulation & World Modeling group builds photorealistic, reactive neural simulators capable of synthesizing counterfactual multi-camera traffic scenarios on demand.

They have engaged Talent247 to hire a Senior ML Engineer specializing in Neural World Simulation & Video Diffusion in London. Their team combines 3D Gaussian Splatting, neural radiance fields, and multi-view latent video diffusion to turn petabytes of fleet driving captures into controllable, closed-loop training environments.

KEY RESPONSIBILITIES

  • Train and scale multi-camera, action-conditioned latent video diffusion and autoregressive world models that maintain geometric cross-view consistency under novel ego-vehicle trajectories.
  • Develop hybrid 3D Gaussian Splatting (3DGS) and generative diffusion pipelines capable of reconstructing dynamic urban scenes and editing weather, lighting, and reactive traffic actors.
  • Optimize inference latency of neural sensor simulators via step distillation, FP8/BF16 mixed precision, and custom Triton/CUDA attention kernels so RL policies can roll out at scale.
  • Engineer automated scene-reconstruction pipelines that process thousands of hours of raw multi-sensor fleet logs with calibrated extrinsics, rolling-shutter compensation, and dynamic object tracks.
  • Design perceptual realism, 3D epipolar consistency, and domain-gap metrics verifying that simulated rollouts correlate tightly with real-road policy behavior.
  • Collaborate closely with RL post-training scientists to synthesize adversarial edge-case scenarios (cut-ins, occlusions, vulnerable road users) for targeted policy hardening.

TECHNICAL ENVIRONMENT & OFFERING

Stack: Python, PyTorch, CUDA/Triton, Diffusers/custom flow-matching frameworks, gsplat / 3DGS / NeRF, Ray, large-scale H100 GPU clusters. Workplace: Hybrid in London, UK. Compensation: £115,000 – £160,000 base salary plus meaningful equity, comprehensive private health coverage, and UK visa sponsorship.

Core Requirements

ESSENTIAL QUALIFICATIONS

  • 4+ years of research engineering or ML engineering experience in 3D computer vision, neural rendering, or generative video modeling (or PhD with strong engineering output).
  • Deep hands-on experience training latent video diffusion, flow-matching, DiT (Diffusion Transformer), or 3D Gaussian Splatting models in PyTorch.
  • Strong grasp of multi-view camera geometry, projective transformations, depth estimation, and spatiotemporal attention architectures.
  • Proven experience scaling distributed multi-GPU training runs and diagnosing throughput, memory, and numerical stability bottlenecks.
  • Clean, modular Python/C++ engineering practices suitable for production-grade simulation infrastructure.

PREFERRED / BONUS EXPERIENCE

  • Prior experience with autonomous driving sensor suites (surround-view cameras, LiDAR, radar) or closed-loop driving simulators.
  • Expertise in diffusion distillation (LCM, DMD, consistency models) or causal streaming video generation.
  • First-author publications or major open-source contributions in CVPR, ICCV, ECCV, SIGGRAPH, or NeurIPS.

Position Telemetry

Location London, UK (Hybrid)
Department Robotics & Embodied AI
Employment type Full-time
Workplace type Hybrid
Salary range
£115k – £160k annual package