About the job
The Model Shaping team at Together AI works on products and research for tailoring open foundation models to downstream applications. We build services that allow machine learning developers to choose the best models for their tasks and further improve these models using domain-specific data. In addition to that, we develop new methods for more efficient model training and evaluation, drawing inspiration from a broad spectrum of ideas across machine learning, natural language processing, and ML systems.
Responsibilities
Design and build Together’s systems and infrastructure for model customization, including user-facing features and internal improvements
Contribute to reliability improvements for the platform, participating in an on-call rotation and improving processes for incident response
Create and improve internal tooling for deployment, continuous integration, and observability
Build a job orchestration platform spanning multiple datacenters, supporting a highly heterogeneous hardware landscape
Partner with teams developing internal services, co-designing these services and incorporating them in systems built within Together
Qualifications
Minimum
3+ years of experience in building infrastructure or backend components of production services
Extensive experience designing, operating, and troubleshooting production Linux environments and Kubernetes-based platforms
Strong software engineering background in Python or Go
Experienced with infrastructure automation tools (Terraform, Ansible), monitoring/observability stacks (Prometheus, Grafana), and CI/CD pipelines (GitHub Actions, ArgoCD)
Cloud environment (e.g., AWS/GCP/Azure) administration experience, preferably with a hybrid bare-metal/cloud environment
Strong communication skills, be willing to document systems and processes and collaborate with peers of varying technical expertise
Comfortable operating across the stack, from cluster operations and infrastructure automation to backend service development
Preferred
Developing large-scale production systems with high reliability requirements
Pipeline orchestration frameworks (e.g., Kubeflow, Argo Workflows, Flyte)
Managing GPU workloads on HPC clusters, ideally with hands-on experience in operating NVIDIA’s networking stack (e.g., NCCL, Mellanox firmware, GPUDirect RDMA)
Deployment of services for AI training or inference
Networking fundamentals, including TCP/IP, DNS, routing, load balancing, TLS, and network debugging tools
Maintaining or contributing to open-source projects