Kubernetes Reliability Architect for AI Model Ops

Obsidian; San Francisco (CA); Remote; USD 120,000 - 160,000; Full time

Job summary

Obsidian seeks a reviewer to evaluate the quality, correctness, and production-readiness of Kubernetes tasks used to train and evaluate frontier AI models. You will assess cluster-operations scenarios, manifest correctness, and failure-mode troubleshooting, providing rubric-based written feedback.

This role requires hands-on Kubernetes expertise in EKS/GKE/AKS or self-managed clusters, with strong scripting in Go, Python, or TypeScript, and ability to document findings clearly for engineering

Qualifications

• 3+ years hands-on production Kubernetes experience (EKS/GKE/AKS or self-managed).
• Deep understanding of cluster internals: CNI/DNS/ingress, PV/PVC storage, RBAC, and failure modes (CrashLoopBackOff, OOMKilled, scheduling/eviction).
• Experience authoring and reviewing manifests / Helm charts and debugging live cluster incidents.
• Proficiency in Go, Python, or TypeScript.

Responsibilities

• Evaluate the quality, correctness, and production-readiness of Kubernetes tasks used to train and evaluate frontier AI models.
• Provide rubric-based written feedback to guide improvements in tooling and workflows.

Skills

Kubernetes

Cluster internals

Manifests/Helm

Go / Python / TS

Tools

Helm

kubectl

Istio

Prometheus

Grafana

EKS/GKE/AKS

Job description

Obsidian seeks a reviewer to evaluate the quality, correctness, and production-readiness of Kubernetes tasks used to train and evaluate frontier AI models. You will assess cluster-operations scenarios, manifest correctness, and failure-mode troubleshooting, providing rubric-based written feedback.

This role requires hands-on Kubernetes expertise in EKS/GKE/AKS or self-managed clusters, with strong scripting in Go, Python, or TypeScript, and ability to document findings clearly for engineering