Skip to main content

Cursor

Software Engineer, ML Infrastructure

San FranciscofulltimemidAdded 1 month ago

About this role

Build and maintain large-scale ML infrastructure for Cursor's coding AI model. Work closely with researchers to optimize training throughput, manage GPU clusters across cloud and bare metal environments, and develop systems for workload scheduling and cluster automation.

What you'll do

  • Collaborate with ML researchers to improve training throughput and reliability
  • Plan and build cutting-edge GPU infrastructure with OEMs and cloud providers
  • Optimize compute density and scalability for large-scale reinforcement learning workloads
  • Develop software and systems to automate GPU cluster building, monitoring, and operations
  • Design workload scheduling and data movement systems for distributed training

What they're looking for

  • Systems and infrastructure software engineering
  • Python, Typescript, Rust, or Golang
  • Distributed storage and networking infrastructure
  • Linux systems administration across cloud and bare metal
  • Infrastructure-as-code and configuration management
  • Kubernetes administration
  • Large-scale distributed systems design
  • GPU operations (Nvidia, Infiniband/RoCE preferred)
Apply with Autofill

Opens the application — the Jobs AI extension fills it for you. Set up autofill

Opens the official application on the employer’s site. No login required.

Cursor

Cursor builds an AI-driven code editor used by millions of developers to transform how software is built. The company is hiring for infrastructure engineers, ML systems specialists, enterprise platform builders, security engineers, and customer success roles focused on driving adoption within large organizations.

Website
cursor.com
View all jobs at Cursor

Likely interview questions

  • Describe your experience building or operating GPU clusters at scale. What were the biggest infrastructure challenges you faced, and how did you solve them?
  • Tell us about a time you optimized training throughput or cluster utilization for ML workloads. What metrics did you focus on, and what was the impact?