Founding AI Accelerator Architect (Research & Hardware Co-Design)

present of AI; Greater London; Hybrid; GBP 120,000 - 190,000; Full time

Job summary

Lamb Labs seeks a Founding AI Accelerator Architect focused on research-to-silicon hardware co-design. You will drive RTL design, performance modeling, and cross-layer decisions to map AI models efficiently into fixed hardware.

The role emphasizes hands-on experience with AI inference optimization and silicon-level architecture, requiring a track record of accelerator concepts reaching silicon. Lamb Labs specializes in custom model processing units for fast inference.

Qualifications

• Deep experience in hardware architecture and AI model co-design.
• Proven track record taking accelerator concepts from research to silicon.
• Experience with RTL design, performance modeling, and cross-domain collaboration.

Responsibilities

• Design and optimize hardware architectures for AI accelerators co-designing with models.
• Lead mapping of neural networks to fixed-function silicon and RTL verification.
• Balance algorithm-hardware tradeoffs and perform cross-layer optimization.

Skills

RTL design

Performance modeling

Model-to-hardware mapping

Cross-layer thinking

Job description

Founding AI Accelerator Architect (Research & Hardware Co-Design)

A founding AI accelerator architect typically needs deep experience in hardware architecture and AI model co-design, with a track record of taking accelerator concepts from research to silicon. Strong candidates bring expertise in RTL design, performance modeling, and the ability to work across algorithm and hardware boundaries. At a company hardcoding model weights into silicon, they must also understand how to map neural network computations efficiently into fixed hardware.

How it fits the AI era

AI is changing accelerator design by making model-specific hardware more valuable, since general purpose chips struggle to keep up with inference demands. This role stays human in its need for creative architecture decisions, tradeoff analysis, and cross-layer thinking that AI tools can assist but not replace. The work also involves anticipating how models evolve, which requires judgment beyond current automation.

How it builds AI experience

This role offers hands on experience with AI inference optimization, model to hardware mapping, and co-design of neural networks with custom silicon. Building such experience can come from projects in FPGA based acceleration, ASIC design for machine learning, or research on efficient inference. Working at a company that hardcodes weights into silicon would deepen understanding of fixed function AI hardware.

Worth showing

• RTL design and verification
• Model to hardware mapping
• Performance modeling
• Silicon tapeout experience

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Lamb Labs

Lamb Labs develops custom model processing units that hardcode artificial intelligence model weights into silicon for fast inference, per the company.