Validation Engineer - AI Infrastructure
No salary listed
Mid
London, UK
HybridHybrid; office visits required.
QA & TestingPerformance Testing, Test Automation Engineering
About the job
Requirements
- Strong proficiency in Python for building zero-dependency verification tools, automated test orchestration harnesses, and low-level system checks.
- Hands-on experience writing and maintaining Infrastructure as Code to provision, configure, and tear down complex bare-metal and containerized compute environments.
- Experience working within Kubernetes environments and validating enterprise-scale, multi-node distributed systems.
- Experience scripting automated checks for server-node topology, PCIe link speeds, HBM memory status, firmware performance, and interfacing with hardware via BMC, IPMI, or Redfish.
- Experience integrating automated infrastructure validation suites into CI/CD pipelines.
- Experience configuring observability pipelines for real-time diagnostic logs and hardware metrics.
- Experience partnering across Engineering, Product, and Datacentre Operations to conduct root-cause analysis on complex platform regressions and hardware-software boundaries.
Responsibilities
- Design and build zero-dependency, Python-based preflight verification tools to validate multi-node distributed initialization, master-to-worker rendezvous routing, and correct GPU-to-CPU process affinity before launching large model-training workloads.
- Write and maintain Infrastructure as Code to provision, configure, test, and tear down complex bare-metal and containerized compute environments.
- Construct adaptive test-orchestration harnesses that detect environment drift and analyse platform changes.
- Automate execution and results aggregation of cluster-level benchmarking suites and high-performance storage benchmarks to validate node-to-node throughput limits.
- Build validation routines to monitor high-throughput network fabrics and evaluate traffic patterns and congestion-control parameters.
- Script low-level automated checks to validate server-node topology, PCIe link speeds, HBM memory status, secure boot parameters, and firmware performance via BMC, IPMI, or Redfish interfaces.
- Integrate automated infrastructure validation suites directly into CI/CD pipelines.
- Configure and maintain observability pipelines to route real-time diagnostic logs and hardware execution metrics to quickly isolate slow, misconfigured, or degrading compute nodes.
- Partner with the Platform team, Network Engineers, and Datacentre Operations to lead root-cause analysis on complex platform regressions, hardware-software boundaries, and distributed interconnect bottlenecks.
Desired Qualifications
- Prior experience validating infrastructure optimized for large model-training workloads, including understanding GPU-to-CPU process affinity and master-to-worker rendezvous routing.
- Understanding of high-throughput network fabrics, traffic patterns, and congestion-control parameters in multi-node GPU clusters.
- Background building and executing cluster-level benchmarking suites and high-performance storage benchmarks to isolate node-to-node throughput limits.
- Experience designing intelligent test systems capable of autonomously detecting environment drift and analysing large-scale platform changes.
About the company
Era4 builds the UK's sovereign AI backbone by converting brownfield sites and industrial hubs into high-performance AI campuses, uniting land, energy, and compute in a controlled system. Its flagship project, The Mill in Newport, transforms a former steel mill into a sovereign UK AI gigafactory with peak power of a gigawatt. The Sovereign AI Mesh spans modular data centres and hyperscale sites that use closed-loop liquid cooling and lower-carbon microgrids. It partners with HPE, VAST Data, WWT, and Civo to deliver domestically controlled AI infrastructure that supports British AI development and prosperity.
Company Size
11-50
Company Stage
N/A
Total Funding
N/A
Headquarters
London, United Kingdom
Founded
2025
Get referred to Era4
See people who can refer or advise you
Simplify's Take
What believers are saying
- On June 9, 2026, Civo and Era4 launched Britain’s first fully sovereign AI stack.
- Derbyshire’s first HPE modular pods target second-half 2026, showing execution momentum.
- Era4 claims forty-plus sites and 50MW capacity, giving it a tangible expansion pipeline.
What critics are saying
- Anglesey County Council rejected Amlwch on April 28, 2026, exposing planning fragility.
- Canterbury’s Shelford Farm plan faced a petition, signaling local resistance to landfill data centers.
- Heavy dependence on Valencia gas sites makes permit delays existential if power deals stall.
What makes Era4 unique
- Era4 pairs UK-only compute with domestic renewable power and zero cross-border data exposure.
- Its landfill-and-brownfield strategy turns stranded industrial sites into distributed AI capacity fast.
- The Civo and HPE partnerships package infrastructure, cloud software, and hardware into one stack.
Help us improve and share your feedback! Did you find this helpful?
Benefits
- Flexible Work Hours