Senior AI Compute & GPU Infrastructure Engineer

Devopsroles; Greater London; Hybride; GBP 120 000 - 150 000; Plein temps

Résumé du poste

Payward (Kraken) is building a dedicated AI Compute and Infrastructure team to power model training, inference, evaluation, and experimentation at scale. You will own GPU clusters, scheduling, and cost-aware infrastructure development across heterogeneous accelerators.

Join a senior, high-impact team working with ML researchers and platform engineers to deliver fast, reliable, and observable AI compute systems in production, with a strong emphasis on uptime and performance.

Qualifications

• 5+ years of infrastructure engineering experience with GPU compute/ML infra.
• Hands-on GPU clusters experience in production environments.
• Strong Linux, networking, storage, containers, Kubernetes fundamentals.
• Experience with ML serving frameworks (vLLM, Triton, TensorRT, TorchServe, KServe, Ray Serve).
• Python proficiency for automation and tooling.
• Understanding of performance tradeoffs: batching, memory, latency, cost.
• Track record of cost optimization while maintaining performance and reliability.
• Experience building observable systems with metrics, logs, dashboards, alerts.
• Clear communicator for researchers, product teams, leadership.
• Comfort in always-on environments with uptime requirements.

Responsabilités

• Own and operate GPU and accelerator clusters for training, inference, evaluation.
• Design infra to run models locally on GPUs to reduce external dependency.
• Build scheduling, orchestration, quota management, and utilization systems.
• Optimize inference pipelines for latency, throughput, memory and cost.
• Collaborate with ML researchers to remove bottlenecks in workflows.
• Build observability for GPU utilization, memory pressure, queue depth, token throughput.
• Drive reliability, incident response, and post-incident improvements.
• Evaluate new hardware and accelerators as the landscape evolves.
• Build tooling to visualize GPU usage for internal teams.
• Contribute to long-term architecture decisions balancing performance and cost.

Connaissances

GPU compute

ML infrastructure

Distributed systems

Linux fundamentals

Kubernetes

Python scripting

Performance optimization

Observability

Cost management

Clear communication

Outils

vLLM

Triton Inference Server

TensorRT

TorchServe

KServe

Ray Serve

DeepSpeed

Megatron-LM

Description du poste

Payward (Kraken) is building a dedicated AI Compute and Infrastructure team to power model training, inference, evaluation, and experimentation at scale. You will own GPU clusters, scheduling, and cost-aware infrastructure development across heterogeneous accelerators.

Join a senior, high-impact team working with ML researchers and platform engineers to deliver fast, reliable, and observable AI compute systems in production, with a strong emphasis on uptime and performance.