Senior Site Reliability Engineer

EPAM is global leader in AI transformation engineering and integrated consulting, serving Forbes Global 2000 companies and ambitious startups. With over thirty years of expertise in custom software, product and platform engineering, we empower our clients to become AI-Native enterprises, driving measurable value from innovation and digital investments.

• 5+ years of experience operating cloud production systems, with a track record in scaling, defining SLOs, managing on-call rotations, and automating manual work away
• Expertise in Azure IaaS/PaaS operations, infrastructure as code (Terraform/Bicep), and CI/CD tooling
• Proficiency in observability stacks with LLM tracing, container orchestration, and Python/Bash automation
• Knowledge of FinOps basics for AI workloads
• Familiarity with daily AI use in operations work, including incident triage, runbook drafting, log analysis, and automation code

• Own deployment end-to-end, including infrastructure as code, CI/CD pipelines, and environment management on Azure, ensuring every environment is rebuildable from source
• Build LLM-aware observability with traces on every model call, production quality signals such as eval sampling, drift detection, and guardrail-trigger rates, plus cost and latency dashboards
• Define and defend SLOs covering availability, latency, and quality objectives per solution, balancing delivery speed against stability with data-driven error budgets
• Run incident management, including on-call models, runbooks written before incidents occur, and blameless postmortems afterward
• Manage the cost of intelligence by monitoring token economics per solution, wiring in budgets and alerts, and conducting capacity planning proactively
• Keep the security posture current through patching, secret rotation, access reviews, and audit readiness across the solution's entire lifecycle
• Shape operability requirements before handover, ensuring they land in the pod's definition of done, and run hypercare jointly with sign-off on what will be operated
• Feed operational patterns, failure modes, and cost learnings back to the pods and the Architect