Staff Site Reliability Engineer at Wikimedia Foundation

As a Staff SRE specializing in ML infrastructure, your primary responsibility is designing, developing, maintaining, and scaling the foundational infrastructure that enables Wikimedia's Machine Learning Engineers and Researchers to efficiently train, deploy, and monitor machine learning models in production.

You will be responsible for:

• Designing and implementing robust ML infrastructure used for training, deployment, monitoring, and scaling of machine learning models.
• Improving reliability, availability, and scalability of ML infrastructure, ensuring smooth and efficient workflows for internal ML engineers and researchers.
• Collaborating closely with ML engineers, product teams, researchers, SREs, and the Wikimedia volunteer community to identify infrastructure requirements, resolve operational issues, and streamline the ML lifecycle.
• Proactively monitoring and optimizing system performance, capacity, and security to maintain high service quality.
• Providing expert guidance and documentation to teams across Wikimedia to effectively utilize the ML infrastructure and best practices.
• Mentoring team members and sharing knowledge on infrastructure management, operational excellence, and reliability engineering.

Skills and Experience:

• 7+ years of experience in Site Reliability Engineering (SRE), DevOps, or infrastructure engineering roles, with substantial exposure to production-grade machine learning systems.
• Proven expertise with on-premises infrastructure for machine learning workloads (e.g., Kubernetes, Docker, GPU acceleration, distributed training systems).
• Strong proficiency with infrastructure automation and configuration management tools (e.g., Terraform, Ansible, Helm, Argo CD).
• Experience implementing observability, monitoring, and logging for ML systems (e.g., Prometheus, Grafana, ELK stack).
• Familiarity with popular Python-based ML frameworks (e.g., PyTorch, TensorFlow, scikit-learn).
• Strong English communication skills and comfort working asynchronously across global teams.