AI Systems Engineer, Codex Agents

Get AI tool; San Francisco, Northern (CA, KY); Hybrid; USD 180,000 - 260,000; Full time

Job summary

OpenAI is seeking engineers to design and build the core agent harness and execution loop for Codex agents in a production setting. You will craft sandboxing, isolation, and workflow infrastructure to support reliable agent orchestration and evaluation across multiple model interfaces.

You will run experiments on prompts, tool-use strategies, and runtime constraints, while improving observability and diagnostics for GPUs and inference systems. Strong collaboration with research is required.

Qualifications

• Experience building or operating production distributed systems, infrastructure, tooling, sandboxing, or ML systems.
• Ability to work across Rust systems code, Python configuration, APIs, and agent orchestration.
• Hands-on experience with LLM applications, coding agents, model deployment, or inference systems.
• Strong focus on reliability, safety, performance, and debuggability.
• Experience collaborating with research while shipping production systems.
• Strong coding skills and ownership of cross-functional AI projects.

Responsibilities

• Design and build the core agent harness and execution loop for Codex agents.
• Build sandboxing, isolation, orchestration, state, and workflow infrastructure for agents.
• Develop evaluation, experimentation, and debugging systems across the agent stack.
• Run experiments on prompts, model interfaces, context construction, and tool-use strategies.
• Improve observability, profiling, and diagnostics across backend systems and GPUs.
• Collaborate with research to make the harness trainable and measurable.
• Build shared platform primitives to improve Codex speed and safety.

Skills

Distributed systems

Rust systems code

Python configuration

Agent orchestration

LLM applications

Reliability and debugging

Research collaboration

Platform engineering

Tools

Rust

Python

Linux

Sandboxing

Cloud platforms

GPU systems

Job description

OpenAI is an artificial intelligence research and deployment company dedicated to building safe artificial general intelligence (AGI) that benefits all of humanity. Founded in December 2015 in San Francisco, it is known for developing systems like ChatGPT and working on advanced AI models.

• Design and build the core agent harness and execution loop for Codex agents.
• Build sandboxing, isolation, orchestration, state, and workflow infrastructure for agents.
• Develop evaluation, experimentation, and debugging systems across the agent stack.
• Run experiments on prompts, model interfaces, context construction, tool-use strategies, and harness behavior to improve reliability, latency, cost, and solve rate.
• Improve observability, profiling, and diagnostics across backend systems, inference, GPUs, and fleet capacity.
• Collaborate with research to make the harness trainable, measurable, and useful for improving frontier AI models.
• Build shared platform primitives that make Codex faster, safer, more reliable, and easier to build upon.

Qualifications

• Experience building or operating production distributed systems, infrastructure, developer tooling, sandboxing, virtualization, cloud platforms, or ML systems.
• Ability to work across Rust systems code, Python configuration layers, APIs, agent orchestration, evaluations, inference behavior, and runtime constraints.
• Hands-on experience with LLM applications, coding agents, model deployment, inference systems, or developer platforms.
• Strong focus on reliability, safety, performance, debuggability, and clean system design.
• Ability to debug ambiguous production failures and deliver durable solutions.
• Experience collaborating closely with research while shipping production systems.
• Strong coding skills and ownership of cross-functional AI systems projects.
• Bonus: Deep experience with Rust, sandboxing, low-level systems, or platform engineering.
• Bonus: Experience with coding agents, agent harnesses, model evaluations, or post-training feedback loops.
• Bonus: Background in compilers, kernels, runtimes, GPU systems, inference optimization, benchmarking, or performance engineering.
• Bonus: Experience building large-scale production infrastructure with high reliability and security requirements.