OpenAI· Codex· Engineering - San Francisco and Seattle
Software Engineer, Codex Cloud
Comp$230K – $325K
Classified Tasks (35)
Automate 0%Augment 86%Human-Only 14%
Augment (30)
AI assists, human decides
Build an AI software engineer (Codex)
technical
Build the Codex agent harness and product iteratively to optimize model-driven software engineering workflows
technical
Train the model iteratively to improve performance on complex software engineering tasks
technical
Build AI systems that write code, reason about software, and act as intelligent agents for developers and non-developers
technical
Own the full lifecycle of experimentation, deployment, and iteration on novel coding capabilities
operational
Build cloud-first software engineering products and supporting platform capabilities to power agentic experiences
technical
Work across the stack to build customer-facing product experiences and fundamental platform capabilities
technical
Own cloud-based agentic experiences such as code review and cloud tasks
operational
Design and build the underlying runtime and orchestration layer for cloud-based agentic tasks
technical
Build and maintain the execution and infrastructure layer that enables Codex to run commands, read and write files, execute tests, and iteratively improve across real codebases
technical
Provide secure, sandboxed runtime environments for Codex to execute operations safely
technical
Design and implement a distributed systems substrate (Codex Runtime) to enable large numbers of supervised AI agents to run in data centers
technical
Identify how teams actually use and break AI-powered software engineering systems
analytical
Drive changes across product, infrastructure, and model behavior to improve Codex reliability for organizations
operational
Design end-to-end customer-facing software engineering experiences across consumer and enterprise segments
creative
Automate engineer workflows to enable safe building and delivery of software to customers
operational
Build container orchestration primitives for enterprise-scale Codex usage
technical
Build virtual machine provisioning and configuration systems for Codex execution environments
technical
Build execution sandboxes for isolated agent operations
technical
Build shared block storage systems for Codex workloads
technical
Build role-based access control (RBAC) systems for Codex access management
technical
Build admin and audit surfaces for platform governance and observability
technical
Build usage metering, rate limits, and pricing control mechanisms for Codex services
technical
Build managed configuration and constraint systems to govern agent behavior and environments
technical
Build analytics that provide teams and operators deep visibility into how Codex is being used
analytical
Design and own secure, observable full-stack systems powering Codex across web, IDEs, CLI, and CI/CD integrations
technical
Integrate Codex systems with enterprise identity and governance systems (SSO/SAML/OIDC, SCIM, policy enforcement)
technical
Build data-access patterns that are performant, compliant, and trustworthy
technical
Use live usage and operational signals to rapidly iterate on product and platform features
analytical
Convert messy, real-world feedback into scalable product and platform improvements
analytical
Human-Only (5)
Requires human judgment
Bring together research, engineering, design, and product to develop the Codex agent harness and product
leadership
Operate across research, engineering, product, and infrastructure to support Codex capabilities
operational
Build the core team responsible for Codex Cloud operations and platform development
leadership
Lead real-world deployments and product launches for Codex systems
leadership
Work directly with customers and go-to-market teams to roll Codex out across customer teams
communication
Job description
Software Engineer, Codex Cloud | OpenAI Careers ## Software Engineer, Codex Cloud Codex - Engineering - San Francisco and Seattle Apply now(opens in a new window) **About the Team** With Codex we’re building an AI software engineer. One that you can pair with, delegate to, or even ask to take on future tasks proactively. Our team is a fast-moving group within OpenAI, bringing together research, engineering, design, and product. We iteratively build the Codex agent harness and product to get the most out of the model, and we iteratively train the model to be great at complex software engineering tasks. The Codex team is responsible for building state-of-the-art AI systems that can write code, reason about software, and act as intelligent agents for developers and non-developers alike. We operate across research, engineering, product, and infrastructure; owning the full lifecycle of experimentation, deployment, and iteration on novel coding capabilities. *Codex Cloud* builds cloud-first software engineering products and the supporting platform capabilities to power agentic experiences across OpenAI. It is a cross-cutting team that works across the stack to build both delightful product experiences and fundamental platform capabilities. **About the Role** Codex Cloud owns cloud-based agentic experiences like code review and cloud tasks (with many more on the way), but also the underlying runtime/orchestration layer for cloud-based agentic tasks. That runtime is the execution and infrastructure layer that turns Codex from a code generator into a real AI software engineer operating at scale. It provides secure, sandboxed environments where Codex can run commands, read and write files, execute tests, and iteratively improve its work across real codebases — just like a human developer. More broadly, Codex Runtime is the distributed systems substrate that enables large numbers of supervised AI agents to run inside OpenAI’s and customers’ data centers, tackling increasingly complex goals over long time horizons. By giving agents a real working context and reliable execution environment, Codex Runtime makes it possible for Codex to validate its own changes, debug failures, and deliver production-ready results, defining how AI agents safely, reliably, and at scale interact with the world’s software. **In this role, you will:** * Shape the evolution of Codex itself by identifying how teams actually use (and break) AI-powered software engineering, and driving changes across product, infrastructure, and model behavior to make Codex a truly reliable teammate for organizations. * Design customer-facing (across consumer and enterprise segments) software engineering experiences end-to-end to automate and empower engineers to safely build and deliver software to their customers. * Build the core team and enterprise primitives that make Codex usable at scale including container orchestration, virtual machine provisioning/configuration, execution sandboxes, shared block storage, RBAC, admin and audit surfaces, usage, rate limits and pricing controls, managed configuration and constraints, and analytics that give teams and operators deep visibility into how Codex is being used. * Design and own secure, observable, full-stack systems that power Codex across web, IDEs, CLI, and CI/CD, integrating with enterprise identity and governance systems (SSO/SAML/OIDC, SCIM, policy enforcement) and building data-access patterns that are performant, compliant, and trustworthy. * Lead real-world deployments and launches by working directly with customers and GTM to roll Codex out across teams, using live usage and operational signals to rapidly iterate and turn messy, real-world feedback into scalable product and platform improvements. **You might thrive in this role if you:** * Have strong software engineering fundamentals and experience turning ideas into production-grade large-scale distributed systems, thinking h