SRE

THE ROLE

As a Site Reliability Engineer at Baseten, you'll define and codify the gold standards of day 2 operations for our ML infrastructure platform. You'll envision and build robust systems, processes, automations, and observability tooling that keep our platform reliable at scale — and that empower the broader organization to operate confidently.

You'll work closely with engineering, forward-deployed and product teams: learning from recurring failure patterns, turning tribal knowledge into automated mitigations, and raising the operational floor for the entire company.

EXAMPLE INITIATIVES

You'll work on projects like these as part of the SRE team

• Improve Baseten SRE Practices, by instrumenting SLOs and SLIs, improving alerting and observability for all services.
• Building AI-assisted tooling for incident triage and response.

RESPONSIBILITIES

• Own the reliability of Baseten's multi-cloud Kubernetes infrastructure, including incident response, post-mortems, and remediation tracking.
• Build and maintain observability infrastructure — metrics, logging, dashboards, and alerting — as code.
• Author, validate, and improve runbooks for recurring failure patterns, ensuring they're structured for low-context, safe execution.
• Identify high-frequency failure patterns and convert them into automated mitigations or self-healing automations.
• Diagnose and resolve runtime issues related to latency, memory behavior, GPU utilization, concurrency, and model lifecycle management.
• Define and instrument SLOs and SLIs across customer workloads and internal services.
• Navigate ambiguity, make principled tradeoffs, and avoid unnecessary complexity in the systems you build and the processes you define.

REQUIREMENTS

• Extensive hands-on experience with Kubernetes (multi-cloud experience across EKS, GKE, or similar is a strong plus).
• Experience in building and maintaining scalable infrastructure.
• Strong foundation in observability tooling: metrics (VictoriaMetrics, Prometheus), logging (Loki, ELK), dashboards (Grafana), and alerting pipelines. Observability-as-code experience is a plus.
• Experience with infrastructure-as-code (Terraform, Helm) and GitOps workflows (Flux CD, ArgoCD).
• Experience writing and improving runbooks, leading incident response, and doing post-mortem analysis.
• Comfort working at the intersection of engineering and operations — you write code, but you also think deeply about process, escalation paths, and operational leverage.
• Familiarity with incident management platforms (incident.io or similar) is a plus.
• No prior ML experience required, but curiosity about how ML models are deployed and served at scale will serve you well.