Pre-Training Engineer
LB347
Posted: 18/11/2026
- $150,000 - $250,000
- Menlo Park, CA
- Permanent
Pre-Training Engineer – Robotics / Multimodal AI / Foundation Models
Menlo Park / San Jose, California, United States (On-site)
I'm working with an early-stage robotics company developing next-generation autonomous manufacturing systems, combining robot learning, computer vision and large-scale AI models.
They're looking for Junior and Senior Pre-Training Engineers with strong deep learning fundamentals and experience training video, vision or multimodal models.
The role:
- Develop and optimise large-scale pre-training pipelines for video and multimodal foundation models.
- Design and train neural networks, including transformers and encoder-decoder architectures.
- Improve training efficiency, scalability and convergence across distributed GPU systems.
- Collaborate with robotics and perception engineers to develop models for real-world applications.
What they're looking for:
- Strong experience in deep learning, neural networks and model training.
- Knowledge of transformers, video models, multimodal learning and numerical optimisation.
- Proficiency in Python, PyTorch and distributed training.
- Strong mathematical fundamentals and ability to implement neural networks from scratch.
- Senior: Experience owning large-scale training runs, debugging failures and optimising performance.
- Junior: Exceptional technical fundamentals, hands-on coding ability and strong research or project experience.
Why consider it?
- Join an early-stage robotics company building next-generation autonomous manufacturing technology.
- Work alongside technical leadership from Tesla Optimus, Figure AI, Neuralink, and Boring Company.
- Own foundational AI models with direct applications in robotics.
- Work across AI research, robotics and GPU computing in a highly technical team.
Location: Menlo Park / San Jose, California – On-site (5 days per week).
Interested? Apply directly or message me for a confidential discussion.
Zee Uddin
Researcher