SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Together AI operates one of the world's largest GPU fleets and is building the software systems that power and automate that infrastructure. The Infra Agent Systems team develops production AI agents that diagnose hardware failures, investigate incidents, correlate signals across the fleet, and automate operational workflows.
In this role, you will work across two interconnected areas:
**Infrastructure Agent Systems**: Build production AI agents that help operate the GPU fleet by diagnosing failures, investigating incidents, gathering evidence from live systems, and assisting with remediation. These agents are used daily by infrastructure and datacenter teams through APIs, CLIs, dashboards, and Slack integrations.
**Core Agent Platform**: Build the foundational platform powering these agents, including knowledge graphs, search and retrieval systems, orchestration frameworks, evaluation tools, and developer tooling that enables agents to reason, act, and continuously improve.
Key responsibilities include designing and building production AI agent systems for infrastructure diagnostics; building distributed services, orchestration frameworks, and knowledge graphs; developing fleet intelligence systems that combine telemetry and operational knowledge; integrating with observability, incident management, and internal systems; owning services end-to-end from architecture through production operations; and improving agent performance through evaluations and feedback loops.
You'll work at the intersection of AI agents, distributed systems, infrastructure, and automation, solving challenging engineering problems with real production impact. This is an opportunity to build foundational systems from the ground up and help define how self-improving AI agents operate real-world AI infrastructure at massive scale.
About Together AI
AI / Data / Infrastructure — cloud platform for open-source and generative AI model training and inference.
