About this role
Your team
You will work with the engineers who connect new machines to Lyceum’s systems and build the orchestration layer that lets us move GPUs between use cases. The team’s aim is to get the most out of every GPU we run. Your part in that: service levels for compute that are clearly defined, monitored and alerted on.
Your mission
You will build the observability stack that shows how every layer of Lyceum's platform is doing, from GPU and memory through Kubernetes to the individual inference worker. Your goal is that we see a problem before our customers do, and that we can warn them before their own workloads fail.
Your focus
Observability platform: Build and run the metrics, logs and traces pipelines across all our data centres, and keep them fast and affordable as the fleet grows.
Alerting and incident tooling: Define SLOs with the product teams, build alerts that are precise enough to act on, and provide the tooling for on-call and incident response.
Proactive detection: Analyse system and customer workload behaviour to catch errors before they occur, and turn recurring incidents into automated checks.
Customer-facing visibility: Give customers a clear view of the usage, performance and health of what they run on Lyceum.
Your KPIs
- Time to detection for incidents
- Share of incidents detected by monitoring before a customer reports them
- Alert precision (actionable alerts versus noise)
- Coverage of platform components with dashboards and SLOs
Your profile
We consider candidates from diverse backgrounds, with a deep love for technical challenges and the desire to take on ownership beyond what's reasonably expected.
Requirements
- 3+ years of experience in site reliability, observability or infrastructure engineering
- Production experience with Grafana, Prometheus, OpenTelemetry or similar tooling
- Strong proficiency in Go and Python
- Strong systems fundamentals: Linux, networking and Kubernetes
- Experience defining SLOs and building alerting that teams trust
Nice to have
- Experience with GPU telemetry and inference serving metrics (time to first token, throughput)
- Experience monitoring multi-cluster or multi-region systems
- Familiarity with model serving frameworks such as vLLM
- Comfortable working on-call or in incident response when things break
Why us?
- Outstanding team: Work with some of the best engineers in the world, coming from hedge funds, big tech, AI startups and top universities
- Once in a lifetime opportunity: Early-stage company in the fastest-growing market in the world
- Ownership: Shape how European AI companies access GPU compute
- European mission: Build sovereign, GDPR-compliant AI infrastructure for the next generation of deep-tech