AI Infrastructure Systems Engineer

Build the infrastructure powering the next generation of AI. At Together AI, you’ll build and operate one of the world’s largest GPU fleets used for frontier model training and inference. This isn’t a traditional infrastructure role—we’re looking for engineers who love building systems, automating everything, and solving problems at massive scale.

Responsibilities

Design and build fleet automation systems that provision, validate, deploy, upgrade, repair, and retire GPU clusters with minimal human intervention; Build AI Infrastructure Agents that automate deployment, root-cause failures, incident triage, and autonomous remediation; Develop Fleet Intelligence platforms that continuously monitor hardware health, firmware, networking, storage, thermals, and workload performance to predict failures before they impact customers; Build software that maximizes GPU availability, utilization, performance, and reliability across thousands of accelerators; Create automated validation systems for GPUs, InfiniBand/RoCE fabrics, NVLink/NVSwitch, storage, and distributed AI workloads; Build internal platforms and developer tools that allow infrastructure to be managed through software—not manual operations; Continuously improve deployment velocity, reliability, and operational efficiency through automation; Partner closely with hardware, networking, platform, and AI teams to push the limits of AI infrastructure

Requirements

3+ years building distributed systems, infrastructure platforms, or large-scale backend software; Strong software engineering skills in Python, Go, or Rust; Experience building platforms, automation systems, or developer infrastructure; Experience with Linux, Kubernetes, Terraform, Ansible, or similar infrastructure technologies; Strong systems thinking with the ability to understand problems across hardware and software; A passion for solving complex infrastructure challenges through software; An automation-first mindset —if a task is repeated, your instinct is to build a system to eliminate it

Nice to have

GPU infrastructure, CUDA, NCCL, NVLink/NVSwitch; InfiniBand or RoCE networking; Bare-metal provisioning and lifecycle management; Large-scale AI training or inference clusters; Hardware health monitoring and predictive failure detection; Distributed storage systems; AI agents and autonomous infrastructure operations

Benefits

startup equity; health insurance; competitive benefits