About this role
Job title: Staff Software Engineer, Model Serving
About the Role As a Staff Engineer at Databricks, you will help shape the Model Serving product and its underlying infrastructure. You will design and build systems that enable high-throughput, low-latency inference across CPU and GPU workloads and influence architectural direction. You will collaborate with platform, product, infrastructure, and research teams to deliver a scalable, governed serving platform.
What You'll Do
- Design and implement core systems and APIs that power Databricks Model Serving, ensuring scalability, reliability, and operational excellence.
- Partner with product and engineering leadership to define the technical roadmap and long-term architecture for serving workloads.
- Drive architectural decisions and trade-offs to optimize performance, throughput, autoscaling, and operational efficiency for CPU and GPU serving workloads.
- Contribute directly to key components across the serving infrastructure — from model container builds and deployment workflows to runtime systems like routing, caching, observability, and intelligent autoscaling — ensuring smooth and efficient operations at scale.
- Collaborate cross-functionally with product, platform, and research teams to translate customer needs into reliable and performant systems.
- Lead technical initiatives that improve latency, availability, and cost-effectiveness across both customer-facing and foundational serving layers.
- Establish best practices for code quality, testing, and operational readiness, and mentor other engineers through design reviews and technical guidance.
- Represent the team in cross-organizational technical discussions and influence Databricks’ broader AI platform strategy.
What We're Looking For
- 10+ years of experience building and operating large-scale distributed systems.
- Deep expertise in model serving, inference systems, and related infrastructure (e.g., routing, scheduling, autoscaling, and observability).
- Strong foundation in algorithms.