AI Systems Engineer, Codex Agents

About this role

Build production-grade AI agent systems at OpenAI that translate model capabilities into reliable, real-world actions. You'll design the core agent harness, execution infrastructure, and evaluation frameworks while working across the full stack from low-level systems to model behavior and GPU fleet optimization.

What you'll do

• Design and build agent harness and execution loops for tool use, code execution, and long-horizon task completion
• Develop sandboxing, orchestration, and state management infrastructure for agents in real development environments
• Create evaluation and debugging systems to isolate issues across harness, model, inference, and product layers
• Run ablations on prompts, model interfaces, and harness behavior to improve solve rate, reliability, latency, and cost
• Improve observability, profiling, and diagnostics across the agent stack from backend to GPU fleet
• Collaborate with research to make the harness trainable and useful for improving frontier models

What they're looking for

• Distributed systems and production infrastructure
• Rust systems programming
• Python and API development
• LLM applications and coding agents
• Debugging and troubleshooting across system layers
• Sandboxing, virtualization, and isolation technologies
• Performance engineering and profiling
• Inference optimization and GPU systems