> Markdown version of [/jobs/ext/4629852-software-engineer-training-infrastructure](https://www.wearedevelopers.com/jobs/ext/4629852-software-engineer-training-infrastructure). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Software Engineer - Training Infrastructure - **Company:** PRESENT PRESENT - **Location:** San Francisco, CA, United States - **Experience:** Expert - **Contract:** Permanent contract - **Skills:** Distributed Computing Environment, Distributed Systems, Machine Learning, Open Source Technology, Reinforcement Learning, Pytorch, DeepSpeed, Kubernetes - **Published:** November 7, 2026 - **Apply:** https://www.juju.com/job/28_25303_f4b8f0ec39c8025cad5f9d786d7c29c0 ## About the Role This role centers on designing and owning the core systems behind a machine learning training platform, including scheduling, storage, networking, reliability, and observability. A strong candidate has at least five years of experience, solid Go skills, and deep hands-on Kubernetes work in production. They also bring a good grasp of distributed systems and tuning, and the judgment to weigh speed against stability while mentoring less experienced engineers. ## Description The work sits close to real model training, including reinforcement learning systems and continuous learning pipelines, built for customers who deploy models on an inference platform. Someone in the role would learn how open source training tools behave at scale and how research engineers use them. To prepare, try running distributed training jobs with PyTorch and a tool such as DeepSpeed or FSDP, and notice where they fail or slow down. Worth showing * Go services in production * Kubernetes at scale * Observability system design * Distributed training experience This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.