About the job
This is a great opportunity to play a defining role in building and operating our inference platform. You will connect deep expertise in distributed systems programming with hands-on knowledge of operating Kubernetes-based platforms according to best practices, all while supporting the platform in production. You will be instrumental in serving cutting-edge machine learning models at massive scale for scientific applications, often requiring you to think from first principles. Success in this role demands excellent technical skills, independence, strong ownership, and a relentless user-focus.
Responsibilities
Contribute to the development and operation of the inference platform, serving fleets of cutting-edge machine learning models to scientific applications.
Deliver high-quality and well-tested user-focused features.
Provide support to users of the platform.
Perform maintenance work and drive internal tech investments for platform stability, reliability and scalability.
Build observability and alerting mechanisms for the platform.
Improve the Continuous Integration/Continuous Deployment (CICD) setup of the platform.
Operate effectively in a fast-paced and ambiguous environment, ensuring independent delivery.
Provide great documentation and guidance for other contributors and users.
Qualifications
Minimum
Experience writing and maintaining Python code in production environments, with an emphasis on concurrent programming (with a strong knowledge of async, threads, processes, GIL, etc).
Experience building, maintaining and operating Kubernetes services.
Experience working with distributed systems.
Experience maintaining APIs that serve a moderately large set of internal users.
Experience working with ML models; an understanding of ML lifecycle and how serving and operating ML models differs from other kinds of workloads.
Preferred
Experience working on an inference platform.
Experience managing a fleet of ML models.
Experience building and maintaining CI/CD processes for complex systems.
Experience with GCP or other comparable clouds.
Experience with building internal and user-focused dashboards.