Data Scientist, Inference Capacity Optimization

About the Role

OpenAI’s Industrial Compute organization is responsible for ensuring our compute infrastructure scales efficiently to support millions of users and increasingly sophisticated AI models.

We’re looking for a Data Scientist to partner closely with Capacity Systems Engineering, Infrastructure, Product, and Research to optimize inference capacity across our global GPU fleet. This role combines statistical modeling, large-scale data analysis, forecasting, and systems thinking to drive critical decisions around infrastructure investments, performance-efficiency trade-offs, and customer experience.

You’ll transform complex operational data into actionable insights that directly influence how OpenAI allocates and scales one of the world’s largest AI compute environments.

Key Responsibilities

• Build statistical and machine learning models to profile and improve GPU utilization, latency, throughput, and overall fleet efficiency.

• Develop forecasting models for inference demand across products, regions, and model families.

• Analyze production workloads to identify latency bottlenecks and capacity constraints, highlighting optimization opportunities.

• Partner with Capacity Systems Engineering to inform infrastructure planning and long-term GPU investment strategies.

• Design experiments and simulations to evaluate scheduling policies, serving strategies, and infrastructure tradeoffs.

• Build dashboards and operational metrics that enable leadership to make data-driven capacity decisions.

• Collaborate with Product, Research, Finance, and Infrastructure teams to align compute planning with business growth and model roadmaps.

• Communicate technical findings clearly to both engineering teams and executive leadership.

Qualifications

• MS or PhD in Statistics, Computer Science, Operations Research, Applied Mathematics, Economics, or related quantitative discipline (or equivalent industry experience).

• 5+ years of experience working in the infrastructure data science space.

• Strong expertise in Python and SQL.

• Experience building forecasting, optimization, or predictive models.

• Strong understanding of experimentation, statistical inference, and causal analysis.

• Experience communicating analytical insights to executive stakeholders.

Preferred Skills

• Capacity planning

• Distributed systems

• AI infrastructure

• Datacenter design and buildout

• Queueing theory

• Time-series forecasting

• Operations research

• Supply-demand modeling

• Reinforcement learning for resource allocation

• Cost optimization