Keith Stevens is a Staff Software Engineer based in Berkeley with 19 years of experience building large-scale AI systems, data infrastructure, and developer-facing tooling. He spent nine years at Google Translate leading a small team to discover, clean, and serve massive web-sourced training datasets and to continuously train and deploy best-in-class translation models. He excels at accelerating small, lean teams—doing the risky, forward-looking architecture and internal tooling that multiplies team velocity while mentoring engineers to ship quickly. Outside Google he’s led engineering at seed-stage AI startups, prototyped MLOps and agentic memory systems, and contributed to prominent open-source AI infra like SkyPilot and OpenAssistant. Known for a playful “Making AI Silly” GitHub persona, he combines deep production ML/infra expertise with hands-on cloud and DevOps work (K8s, multi-cloud, encrypted AWS volumes) that reliably moves research into repeatable production.
OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.
Role in this project:
Back-end & Front-end Developer
Contributions:521 reviews, 363 commits, 301 PRs in 2 months
Contributions summary:Keith primarily worked on developing a Discord bot for the OpenAssistant project, adding features like slash commands and user registration. They implemented backend interaction using Python and utilized the Discord API. In addition, they contributed to the project's website by integrating Next.js and Supabase for authentication and UI elements.
SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 16+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
Role in this project:
Cloud Engineer / DevOps Engineer
Contributions:15 reviews, 6 PRs, 39 comments in 1 year
Contributions summary:Keith primarily contributed to infrastructure and cloud-related aspects of the SkyPilot project. Their work included adding support for environment variables and setting up testing related to it, which indicates a focus on the configuration and deployment process. They also implemented features for AWS, such as enabling encrypted disks, and propagating Kubernetes custom metadata. Further work included returning service names and endpoints for the `sky.serve.up` method, and various testing and unit testing improvements.
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.