About the job

OLIX is seeking an AI Performance Modeller to design, build, and own functional and high-fidelity, cycle-accurate performance models that guide the evolution of its optical compute architecture and software stack. The role sits at the intersection of hardware architecture, software tooling, and machine-learning workload analysis, focusing on building simulators for OTPU subsystems and full pipelines, identifying latency, throughput, and utilisation bottlenecks, and driving data-backed architectural and scheduling improvements. The candidate will analyse ML workloads such as LLMs, diffusion models, and graph workloads, run large-scale design-space explorations, and translate complex simulation results into clear, quantitative recommendations for hardware, software, and optical teams. The role involves developing robust Python and C++ tooling for trace parsing, statistical analysis, visualisation, and CI integration, ensuring performance validation on every design iteration. OLIX offers a highly competitive salary, equity, full relocation support, UK visa sponsorship, premium healthcare benefits, and a collaborative on-site working environment in London.