Applied AI Engineer for Codex Core Agents

OpenAI; San Francisco (CA); On-site; USD 100,000 - 140,000; Full time

Job summary

A leading AI research organization in San Francisco seeks applied AI engineers to enhance Codex agents for software engineering tasks. This role involves designing agent behaviors, analyzing production failures, and collaborating with research teams on performance evaluation. Candidates should have a strong background in Python and experience with machine learning products. The company values diverse perspectives and is an equal opportunity employer. This position offers the chance to transform research into practical applications. Targeting candidates interested in turning model potential into user-friendly systems.

Qualifications

• Experience building or shipping machine learning or LLM-powered products.
• Strong in Python and comfortable with modern ML tooling.
• Experience with model evaluation, fine-tuning, or prompt design.
• Ability to think in terms of systems and user outcomes rather than just model metrics.
• Enjoy debugging messy, real-world failures.

Responsibilities

• Design and iterate on agent behaviors across real-world coding tasks.
• Work closely with research to measure agent performance.
• Improve performance through various strategies.
• Analyze failures in production and systematically improve reliability.
• Build feedback loops that enhance evaluation and research.

Skills

Python

Machine learning

LLM-powered products

Debugging

System thinking

Job description

A leading AI research organization in San Francisco seeks applied AI engineers to enhance Codex agents for software engineering tasks. This role involves designing agent behaviors, analyzing production failures, and collaborating with research teams on performance evaluation. Candidates should have a strong background in Python and experience with machine learning products. The company values diverse perspectives and is an equal opportunity employer. This position offers the chance to transform research into practical applications. Targeting candidates interested in turning model potential into user-friendly systems.