About the role

Role Overview

Join Baseten as a Software Engineer on the Training Infrastructure team. You will architect and lead the development of Baseten’s training platform, making key infrastructure technical decisions that help developers deploy, scale, and monitor workloads reliably.

Responsibilities

Architect and lead the training platform’s infrastructure and technical direction.
Design scalable infrastructure for ML training, including:
- Scheduling
- Storage
- Networking
Partner closely with developers and research engineers.
Drive long-term improvements in:
- System reliability
- Development velocity
- Technical strategy
Develop/own observability systems for ML infrastructure.
Improve distributed performance and reliability across training workloads.

Requirements

Proven experience designing observability systems.
Bachelor’s degree or higher in Computer Science (or related field).
Advanced distributed systems knowledge and performance tuning.
Deep expertise with Kubernetes in production.
Proficiency in Go (with Python experience as a plus).
Extensive experience with major cloud providers (AWS, GCP).
Experience with distributed storage systems.
Experience with workload orchestration platforms such as Temporal or Airflow.

Nice to Haves

Experience with ML/AI workloads and MLOps platforms.
Experience with additional cloud/compute providers (e.g., Crusoe, DigitalOcean, Nebius).
Familiarity with open-source training stack/frameworks (e.g., NCCL, PyTorch, Megatron, NemoRL, VeRL, Axolotl, HF Trainer).
Experience with distributed training techniques (FSDP, DeepSpeed).
Experience building AI products, tooling, or agents.

About Baseten

Baseten is a remote-first company focused on building infrastructure and platforms for machine learning developers. Its Training Infrastructure team works on scalable systems that enable efficient deployment, scaling, and monitoring of ML workloads with high performance and reliability.