Senior Machine Learning Engineer, DevOps​/SRE

Teamwork makes the stream work.
Roku is changing how the world watches TV

Roku is the #1 TV streaming platform in the U.S., Canada, and Mexico, and we've set our sights on powering every television in the world. Roku pioneered streaming to the TV. Our mission is to be the TV streaming platform that connects the entire TV ecosystem. We connect consumers to the content they love, enable content publishers to build and monetize large audiences, and provide advertisers unique capabilities to engage consumers.

From your first day at Roku, you'll make a valuable - and valued - contribution. We're a fast-growing public company where no one is a bystander. We offer you the opportunity to delight millions of TV streamers around the world while gaining meaningful experience across a variety of disciplines.

The Advertising Performance group focuses on performance for all participants in the Advertising ecosystem
- Advertisers, Publishers, and Roku. The systems and solutions span multiple disciplines and technologies to perform real-time multi-objective optimization across distributed systems at large scale and with low latency. We use Machine Learning, Reinforcement Learning, AI, Control and Optimization Systems, and Auction Dynamics to solve a large set of complex problems. At the core of this is our Machine Learning, Experimentation, and Inference Platform that powers the entire landscape, which we continuously evolve over time.

We are seeking a talented and experienced Senior Software Engineer, MLOps/Dev Ops, to join the Advertising Performance team and play a critical role in supporting and scaling our Machine Learning infrastructure. The ideal candidate has a strong background in Dev Ops/SRE practices, cloud infrastructure management, and MLOps tooling — with a passion for building platforms that accelerate ML experimentation and deployment at internet scale.

You will partner closely with ML Scientists and Engineers to streamline the end-to-end ML lifecycle across training, evaluation, deployment, and monitoring — on top of a modern, cloud-native stack running on GCP and AWS using Kubernetes, Apache Airflow, Spark, Ray, MLflow, Chronon, etc.

For California Only
- The estimated annual salary for this position is between $148,750 - $361,000 annually. Compensation packages are based on factors unique to each candidate, including but not limited to skill set, certifications, and specific geographical location. This role is eligible for health insurance, equity awards, life insurance, disability benefits, parental leave, wellness benefits, and paid time off.

• Lead the design and operation of scalable, production-grade cloud infrastructure for ML workloads across AWS and GCP, including GPU/TPU-based training and inference environments
• Architect and improve CI/CD systems for ML models and platform services to enable fast, reliable, and safe production releases
• Own and evolve low-latency infrastructure for real-time model inference, including KV store and vector databases
• Define and enforce observability standards for ML systems, including model performance monitoring, drift detection, capacity planning, and pipeline health metrics
• Participate in on-call rotation, leading incident response and root-cause analysis for critical ML training and serving infrastructure
• Partner with data scientists and ML engineers to improve platform usability, accelerate model iteration, and implement strong MLOps and SRE best practices
• Champion operational excellence across ML infrastructure through automation, resilience engineering, disaster recovery planning, and continuous improvement

• BS or MS in Computer Science, Engineering, or a related quantitative field
• 8+ years of experience in Dev Ops, SRE, or ML infrastructure, including 4+ years supporting large-scale ML or AI systems
• Strong programming skills in Python, and/or Scala, or Java for platform automation and tooling
• Deep experience with Kubernetes and container orchestration on GCP (GKE) and/or AWS (EKS)
• Expertise with No
  
  SQL or low-latency data stores such as Aerospike or similar technologies
• Hands‑on experience…