Data Scientist, Core Experimentation
OpenAIAbout the Team
The Statsig team at OpenAI builds and operates the experimentation platform that powers product development, measurement, and decision-making across the company. We partner closely with product, engineering, and infrastructure teams to ensure experiments are trustworthy, statistically rigorous, and scalable to the needs of frontier AI products.
Our mission is to help teams make better decisions through reliable experimentation. We care deeply about statistical correctness, pragmatic solutions, and building systems that researchers and engineers can trust at massive scale. The team operates at the intersection of experimentation methodology, data infrastructure, causal inference, and product analytics.
We are looking for experienced experimentation experts who want to shape the future of experimentation in the AI era.
About the Role
We are hiring a Staff-level Data Scientist to help lead the evolution of OpenAI’s core experimentation platform. This role is focused on improving the statistical rigor, reliability, and practical usability of experimentation across the company.
You’ll work on some of the hardest problems in online experimentation: sample ratio mismatch detection, variance reduction, bias mitigation, metric design, triggered analysis, heterogeneous treatment effects, sequential testing, and experimentation in complex ML systems. You’ll also help translate advanced statistical concepts into pragmatic systems and product experiences that teams can actually use.
This is a highly technical individual contributor role with significant influence across methodology, platform architecture, and experimentation best practices. The ideal candidate combines deep statistical expertise with strong systems intuition and hands-on experience building or operating experimentation platforms at scale.
In this role, you will:
Drive the statistical direction and technical strategy for OpenAI’s experimentation platform
Design and improve experimentation methodologies used across product and research teams
Build pragmatic solutions to real-world experimentation challenges, balancing rigor with operational simplicity
Improve the reliability and trustworthiness of experiment results, including detection and prevention of bias, logging issues, and data quality failures
Developscalable analytica