Staff ML Ops Engineer

📋 Description

• • Own the end-to-end lifecycle of Calix’s ML and GenAI infrastructure, ensuring every model we ship is reliable, scalable, and cost-efficient from first experiment to millions of daily predictions.
• • Architect, build, and continuously improve cloud-native platforms on Google Cloud that power distributed training, fine-tuning, and low-latency inference for state-of-the-art broadband analytics, customer-experience AI, and generative applications.
• • Design Kubernetes-based orchestration layers that abstract CPU, GPU, TPU, and emerging NPU resources so data scientists can focus on breakthrough algorithms while the platform transparently handles placement, autoscaling, and fault tolerance.
• • Implement hardened CI/CD pipelines (GitOps, automated testing, progressive delivery) that shrink model release cycles from weeks to hours and guarantee zero-downtime rollouts across multi-region clusters.
• • Establish and evangelize MLOps best practices—experiment tracking with MLflow, feature stores, model registry, data & concept-drift monitoring—raising the bar for reproducibility, governance, and compliance across the entire AI organization.
• • Partner daily with data scientists, ML engineers, and product teams to translate research prototypes into hardened micro-services, advising on everything from feature engineering to latency budgets and cost modeling.
• • Optimize every layer of the stack: container images, distributed training strategies, model quantization, caching, and autoscaling policies that together cut cloud spend while doubling throughput for peak traffic events.
• • Build comprehensive observability—metrics, traces, logs, and alerting—using Prometheus, Grafana, and the ELK stack so anomalies are caught and resolved before customers notice.
• • Champion security and privacy by design: IAM policies, network isolation, secrets management, and data encryption that keep subscriber data safe and satisfy SOC 2 and ISO 27001 auditors.
• • Contribute to Calix’s open-source initiatives and internal guilds, sharing reusable Terraform modules, Helm charts, and Python/Go SDKs that accelerate every team’s journey to production-grade AI.
• • Mentor junior engineers and create runbooks, design docs, and brown-bag sessions that scale your expertise across the company.
• • Stay on the leading edge—evaluate new runtimes (KubeFlow Pipelines v2, Ray Serve, Triton Inference Server) and emerging hardware (Cloud TPU v5, NVIDIA H100) to keep Calix two steps ahead of the broadband industry’s AI curve.