VLA Pre-training & World Models | Humanoid

Hirehire

London 10/21/2026 Full Time

Role: VLA Pre-training Engineer / Deep Learning Engineer – World Models (2 openings) Company: <a target="_blank" rel="noopener noreferrer nofollow" href="https://thehumanoid.ai/">Humanoid</a> Location: London, UK (King’s Cross) – on-site, five days a week Job Type: Full-timeAbout HumanoidHumanoid is building the world’s most capable, commercially scalable and safe humanoid robots. Its platform, HMND-01, is already deployed in real industrial environments, which makes Humanoid the most serious humanoid robotics company in Europe.<ul style="min-height:1.5em"><li>300+ people, 34+ nationalities, flat structure: your decisions ship directly into real robots.</li><li>Offices in London (HQ, King’s Cross), Boston, Vancouver and San Diego.</li><li>Values: Speed, Ambition, Passion, Trust.</li></ul>About the roleThe VLA Core team builds the base vision-language-action model that the entire manipulation pipeline depends on. This is deep learning at scale: training models from scratch on multi-embodiment trajectory corpora and internet-scale video, with real deployment constraints. The model has to fit and run on a real robot.Robotics experience is not required. If you come from LLMs, VLMs or video generation, be ready to get up to speed on a new domain quickly.This is a senior individual contributor role with no management track. The team is research-adjacent, but the output is models that run on robots. Results are shared as blog posts, not papers.There are two tracks in one team. They are the same role with a different focus, so you can lean into either.Track 1: VLA Pre-training<ul style="min-height:1.5em"><li>Pre-train base models on a diverse multi-embodiment corpus of trajectories, fine-tune policies to excel at specific tasks and explore synthetic data.</li><li>Post-train policies with behaviour cloning and RL, and own the full loop from data to deployment.</li><li>Partner with the Data Collection team: specify what good data looks like, identify failure modes, ensure diversity and coverage.</li><li>Run pre-, mid- and post-training on the VLA stack; explore new modalities and architecture changes.</li><li>Build and maintain continuous pipelines: ingest synthetic data and teleop logs, version them, apply weak-supervision labelling, curate balanced datasets and automatically surface fresh failure cases into retraining.</li><li>Work with the MLOps and Data Platform teams to scale distributed training and optimize models for real-time edge inference on the robot.</li></ul>Track 2: World Models<ul style="min-height:1.5em"><li>Design and train multimodal, action-conditioned world models (video, state, action and language) using diffusion-based and transformer architectures.</li><li>Build video prediction and dynamics models that stay physically consistent over long horizons, including contact-rich manipulation, and serve as pretrained priors for VLA policies.</li><li>Develop learned-simulator evaluation: score candidate policies offline, predict real-world success rates before deployment, roll out policy futures for safety review and planning.</li><li>Generate synthetic rollouts and counterfactual experience (rare events, cross-platform and sim-to-real transfer) and measure their effect on downstream task performance.</li><li>Build data pipelines that turn fleet telemetry, teleoperation logs and internet-scale video into training corpora.</li><li>Run scaling and ablation studies on architecture, data mixture and context length.</li></ul>What you need to succeed<ul style="min-height:1.5em"><li>3+ years building deep learning systems in industry or research, with shipped models or published work to show for it.</li><li>Deep hands-on experience with at least one of: LLMs, VLMs, image or video generative models (architecture, training and inference). For the World Models track: diffusion models, autoregressive transformers, latent-variable models or video prediction.</li><li>You have actually trained models yourself: pre-training runs, distributed infrastructure, the works. Connecting APIs, building agents or a quick fine-tune is not enough.</li><li>Experience with deep learning infrastructure: streaming datasets, checkpointing and state management, distributed training strategies, debugging numerics and training instabilities.</li><li>Strong Python and PyTorch or JAX. You can profile, optimize data loaders and write maintainable research code.</li><li>Strong coding fundamentals. The process includes a LeetCode-style coding round, so come prepared.</li><li>Empirical rigor: careful evaluations, honest baselines, clear experiment docs, crisp communication of trade-offs.</li><li>You are proactive: you can take an open, messy problem and drive it without waiting for a spec.</li><li>Working English.</li></ul>Nice to have<ul style="min-height:1.5em"><li>Robotics or autonomous driving experience; familiarity with simulators (Isaac Sim, MuJoCo) and sim-to-real.</li><li>Experience applying RL to LLMs or robotics.</li><li>Experience with VLA models; familiarity with OpenVLA, Physical Intelligence (π) models or similar open-source frameworks.</li><li>Experience with world models for robotics or autonomous driving (action-conditioned video models, learned simulators, model-based RL).</li><li>Proven productization of deep nets: latency and throughput constraints, on-device optimization, fast inference for generative models.</li><li>Publications at top-tier DL conferences (NeurIPS, ICML, ICLR, CoRL, CVPR) or equivalent open-source contributions.</li></ul>Interview process<ul style="min-height:1.5em"><li>Intro call with Igor, hirehire’s AI researcher.</li><li>HR screen with Humanoid.</li><li>Coding round (LeetCode-style).</li><li>Specialization interview (pre-training or world models).</li><li>Technical execution interview: how you work, collaborate and drive problems.</li><li>Optional chat with a tech leader.</li></ul>What you get<ul style="min-height:1.5em"><li>Salary: a wide range, set by the level you show in the interviews.</li><li>Equity: stock options with meaningful upside as Humanoid scales.</li><li>30+ paid days off: 23 days annual leave, all UK bank holidays and a company shutdown between Christmas and New Year.</li><li>Private healthcare: fully funded, including virtual and in-person care.</li><li>Pension: 8% total contribution (5% employee, 3% employer) on full earnings.</li><li>Daily food: free breakfast, catered lunch and snacks in the office.</li><li>Real ownership: direct access to founding leadership, real input on product direction and room to drive key initiatives from day one.</li><li>Visa sponsorship and relocation to London.</li></ul>Find more <a href="https://www.arbeitnow.co.uk/english-speaking-jobs">English Speaking Jobs in United Kingdom</a> on Arbeitnow</a>

AI/ML/Tech

VLA Pre-training & World Models | Humanoid

Related on GOH