Neural Network Performance Engineer
HumanoidVisit Humanoid website
UK-based AI and robotics company building commercially scalable humanoid robots for industrial deployment.
London, United Kingdom
Funding history
About Humanoid
Humanoid develops industrial humanoid robots and its proprietary KinetIQ AI framework for real-world tasks across manufacturing, logistics, and related environments.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will analyze model performance bottlenecks and improve inference efficiency across cloud and onboard hardware. You will adapt models to new hardware, implement custom kernels, quantize models, and evolve model architectures while preserving their capabilities.
Requirements
- 3+ years building deep-learning systems
- 1+ years optimizing neural-network inference performance
- GPU architecture
- Python
- PyTorch or JAX
- Profiling
- Numerical debugging
- Research-code development
- Experiment documentation
Responsibilities
- Analyze model-architecture performance bottlenecks
- Adapt models to new hardware efficiently
- Implement custom kernels to reduce memory-throughput requirements
- Quantize models with minimal quality loss
- Implement architecture changes that improve performance without sacrificing capabilities
Benefits
- Stock options
- 30+ paid days off including 23 days of annual leave, UK bank holidays, and company closure days
- Private healthcare
- Pension scheme with 8% total contribution
- Free daily breakfast, catered lunch, and snacks in-office