Software Engineer, Safety Engineering

OpenAI • San Francisco, CA

6d • $185,000 - $385,000

About The Position

The Safety Systems organization at OpenAI is responsible for ensuring the safe deployment of AI models to benefit society, aligning with OpenAI's mission to build and deploy safe Artificial General Intelligence (AGI). The Safety Engineering team specifically focuses on building the platforms and tools necessary to make OpenAI's models safe for real-world use. This involves close collaboration with researchers, product teams, and policy experts to translate safety concepts into scalable and reliable systems. Their work encompasses risk measurement, safeguard enforcement, and continuous improvement of model behavior in production, operating at the intersection of product engineering, data, and AI. The role involves building full-stack internal tools to enhance the safety and reliability of OpenAI's models, with a particular emphasis on sensitive areas like mental health and protections for vulnerable users. The goal is to increase the team's efficiency in identifying and resolving safety issues and to shorten the feedback loop between policy, data, and model training. The position also includes the potential to transform successful AI-assisted safety workflows into external-facing products that promote safer AI development across the industry and prepare for more capable AGI.

Requirements

• 5+ years of relevant engineering experience at tech and product-driven companies
• Proficient with JavaScript, React, and other web technologies
• Proficient with at least one backend language (we use Python)
• Experience with relational databases like Postgres/MySQL
• An interest in AI/ML (direct experience not required)
• Excited to partner closely with researchers and policy writers to ship tools that directly improve the safety of OpenAI’s models

Responsibilities

• Own the end-to-end development of internal tools that help improve the safety of OpenAI’s models (with a focus on areas like mental health and other vulnerable-user protections)
• Partner closely with Safety Systems researchers, engineers, and model policy creators to understand workflows, pain points, and requirements—and translate them into durable product solutions
• Build full-stack experiences to support core model policy workflows, such as labeling and inspecting data, analyzing and reviewing failure cases, and surfacing insights for iteration
• Optimize internal applications for usability, speed, and scale to increase team velocity and reduce time-to-fix for safety issues
• Transform successful AI assisted safety workflows into external facing safety products, that empowers developers to build safer AI, uplift the industry standard for AI safety, and prepare the world for more capable AGI.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed