Sr. Site Reliability Engineer

Role Overview

We are seeking a high-caliber Site Reliability Engineer (SRE) to join our Forward Engineering team. You will be the guardian of our production ecosystems, ensuring that our complex, data-driven AI platforms remain resilient, scalable, and highly performant. This role is a hybrid of software engineering and systems architecture, with a specialized focus on MLOps—bridging the gap between model development and production-grade reliability.

Key Responsibilities

1. Reliability & Performance Engineering

• SLA/SLO Management: Define, monitor, and maintain Service Level Objectives (SLOs) and Service Level Indicators (SLIs) for critical AI/ML services.
• Error Budgeting: Manage error budgets to balance the velocity of feature releases from the ML team with the stability of the production environment.
• Scalability: Architect and manage auto-scaling strategies for Kubernetes (GKE) to handle fluctuating workloads during model training and high-volume inference.

2. MLOps & AI Infrastructure

• Model Serving Reliability: Ensure the high availability of Vertex AI endpoints and custom inference services.
• GPU/TPU Optimization: Monitor and optimize compute resource utilization (accelerators) to ensure cost-efficient performance for Large Language Models (LLMs).
• Pipeline Resilience: Support and stabilize ML pipelines (Vertex AI Pipelines/Kubeflow) to ensure seamless data flow from ingestion to model retraining.

3. Automation & Orchestration (Eliminating "Toil")

• Infrastructure as Code (IaC): Use Terraform or Pulumi to provision and manage consistent, version-controlled cloud environments.
• CI/CD & GitOps: Design and optimize robust deployment pipelines for both application code and ML models using GitHub Actions, Cloud Build, or ArgoCD.
• Task Automation: Develop custom Python or Go scripts to automate repetitive operational tasks, self-healing mechanisms, and resource cleanup.

4. Monitoring, Alerting & Incident Response

• Observability: Build and manage comprehensive dashboards using Prometheus, Grafana, or Google Cloud Operations Suite (Stackdriver).
• Incident Management: Act as a primary responder in on-call rotations, leading the technical resolution of production outages.
• Blameless Post-Mortems: Conduct deep-dive root cause analysis (RCA) to ensure systemic issues are identified and permanently remediated through code.