Senior Performance Engineer | AI Infrastructure | Cambridge (Hybrid) |

Overview

In this role you will bridge research and engineering to turn live training and inference data into actionable performance insights. You’ll build models and calculators to forecast hardware and architecture value, guiding purchases and system design across the organization and its members. The work shapes real-world AI infrastructure decisions at a well-funded Cambridge non-profit striving to reduce duplication of effort. This is a rare early-stage opportunity with genuine backing and broad cross‑functional impact.

Pay / Benefits

• Competitive salary and pension
• Hybrid work model from Cambridge office
• Exposure to AI and academic community
• Backing and momentum behind the project

Responsibilities

• Extract real-number signals from live training and inference runs
• Build performance models and calculators to forecast system behavior and ROI of optimisations
• Evaluate whether different accelerators or architectural changes pay off
• Influence purchasing decisions and system design across the org and member institutions
• Collaborate between research and engineering teams to turn data into actionable insights

Key requirements

• A degree in computer science, mathematics, or adjacent field
• Track record of building performance models or calculators (Python or spreadsheet-based) to forecast system behavior
• Hands-on GPU/accelerator code optimisation (CUDA or similar)
• Solid understanding of how LLMs and DL models run on hardware (training vs inference, matrix multiplications, KV-caching)
• Experience with profiling tools: Nsight or PyTorch Profiler
• Experience with monitoring stacks: Prometheus and Grafana
• Python for data work: Pandas and NumPy, plus scripting

• cross-functional collaboration
• analytical mindset
• problem solving
• CUDA or similar GPU/accelerator optimization
• LLMs and deep learning hardware knowledge
• Profiling with Nsight or PyTorch Profiler

…