P-1096

At Databricks, we run the world's largest AI platform. We believe the next breakthroughs will come from models trained on domain-specific enterprise data, not from bigger general-purpose models alone.

That's why we built AI Runtime, a serverless product that makes GPU training of any size feel simple, so companies can train their own models the way frontier labs do. Customers across industries train on it, right next to the enterprise data already in Databricks. Among them are Rivian in automotive, Robinhood in fintech, and Merck, which builds foundation models for drug discovery. Our own AI researchers use it to post-train models with hundreds of billions to nearly a trillion parameters.

It takes a lot of engineering to make that work, and we've written about it on our blog: how we keep GPUs reliable, make training fault-tolerant, and speed up fine-tuning with fused kernels and smart sequence packing. AI agents now launch experiments too, and on our research team they've already increased training jobs 8x.

As a Senior Staff Software Engineer, you'll work on the systems behind all of this, solving the hardest technical problems and shaping where AI Runtime goes next.

The impact you will have:

• Drive AI Runtime's architecture and lead Databricks' long-term technical direction in AI training infrastructure.
• Define the reliability and observability strategy for a fleet that keeps growing in scale and complexity.
• Shape AI Runtime's developer experience for people and agents, so even the largest training jobs feel as simple as a first experiment.
• Anticipate industry trends and 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 decide what we build next and align teams on hard technical calls.
• Champion engineering excellence, raise the bar, and help us hire and grow great engineers.

What we look for:

• 15+ years in large-scale distributed systems, with significant depth in GPU training, high-performance computing, or ML systems.
• Hands-on experience training large models with FSDP, DeepSpeed, or Megatron, and fluency in parallelism strategies.
• Expertise in GPU performance, including accelerator architecture, high-speed interconnects, and kernels.
• Strong understanding of training resilience patterns, such as checkpointing, failure detection, and automatic recovery.
• A background in building and running multi-tenant cloud platforms as products.
• A track record of setting technical direction across teams and seeing it through to business impact.
• The credibility to bring teams to agreement on difficult trade-offs, and a reputation as a mentor senior engineers seek out.
• BS in Computer Science or a related field (MS or PhD preferred).