仕事内容
<h3><strong>About the Role</strong></h3>
<p>Together AI runs one of the most demanding GPU fleets in the industry. Keeping that fleet healthy - every node online, every GPU performing, every datacenter transition running on schedule - is operationally complex and genuinely high-stakes. We're looking for a Junior TPM to own that operational reality.</p>
<p>This is not a coordination or status-reporting role. You will own the end-to-end node lifecycle - from the moment a node goes down through repair, return, and re-integration - and you'll drive the cross-functional work to close every gap as fast as possible. You'll manage datacenter bring-ups, hunt down GPU utilization loss, and build the processes and dashboards that make our fleet operations more visible and accountable over time.</p>
<p>The environment moves fast and doesn't always come with a clear playbook. Much of what you'll work on is genuinely novel - you'll be figuring things out alongside engineers who are building at the frontier. If that sounds like an obstacle, this isn't the right role. If it sounds like the best possible way to learn, keep reading.</p>
<h3><strong>Responsibilities</strong></h3>
<ul>
<li>Own the end-to-end node lifecycle - from failure through repair, return, and re-integration — across provider ticketing, internal tooling, and the state machine that governs each stage</li>
<li>Drive node remediation to resolution with urgency, eliminating gaps in ownership at every handoff</li>
<li>Manage project timelines for new datacenter bring-ups, coordinating across internal teams and external providers to keep milestones on track</li>
<li>Identify and diagnose GPU utilization loss across the fleet, working with engineering leads to drive resolution</li>
<li>Build dashboards and tracking processes that make efficiency gaps visible and ensure they get closed</li>
<li>Continuously improve operational workflows through process improvements and lightweight automation</li>
<li>Develop and maintain