Staff AI Runtime Systems Engineer (GPU Training)

Fdetimes; Mountain View (CA); On-site; USD 190,000 - 265,000; Full time

Job summary

Databricks is seeking a Staff Software Engineer to own major parts of AI Runtime, the serverless platform for GPU training. You will shape architecture, improve reliability, and advance observability across a growing fleet of training jobs.

You will mentor engineers, collaborate with research and product teams, and prepare the platform for next-generation GPUs and workloads, balancing performance, scalability, and developer experience.

Qualifications

• 10+ years in large-scale distributed systems with GPU training or ML systems.
• Hands-on experience training large models with FSDP, DeepSpeed, or Megatron.
• Fluency in parallelism strategies and GPU performance optimization.
• Experience building multi-tenant cloud platforms as products.
• Strong mentoring and cross-team leadership.

Responsibilities

• Drive major parts of AI Runtime's architecture and help shape Databricks' long-term technical direction in AI training infrastructure.
• Design the reliability and observability systems for a fleet that keeps growing in scale and complexity.
• Lead improvements to AI Runtime's developer experience for people and agents, so even the largest training jobs feel as simple as a first experiment.
• Help choose the technical bets that prepare AI Runtime for new GPU generations and workload types.
• Take on the hardest multi-node training problems yourself, from orchestration and GPU scheduling to data loading and checkpointing.
• Make training faster and cheaper by raising efficiency across model architectures and GPU accelerators.
• Work with research, product, and customers to shape what we build next and lead projects across teams.
• Champion engineering excellence, raise the bar, and mentor the engineers around you.

Skills

Distributed systems

GPU training

High-performance computing

ML systems

Mentoring

Education

BS in Computer Science or related field

Tools

FSDP

DeepSpeed

Megatron

Job description

Databricks is seeking a Staff Software Engineer to own major parts of AI Runtime, the serverless platform for GPU training. You will shape architecture, improve reliability, and advance observability across a growing fleet of training jobs.

You will mentor engineers, collaborate with research and product teams, and prepare the platform for next-generation GPUs and workloads, balancing performance, scalability, and developer experience.