Infrastructure Capacity Planner - Flexible Hours & Hybrid

Anthropic; San Francisco (CA); On-site; USD 320,000 - 405,000; Full time

Job summary

Anthropic is seeking an Infrastructure Capacity Planner to own medium-range demand forecasts for accelerators, CPU, storage, network, and managed services. You will interface with finance for cost efficiency and partner with data teams to refine planning tools and telemetry.

The role focuses on model-building using SQL and Python, with an emphasis on transparent forecasting and workload optimization. Based in San Francisco, some hybrid office time is expected.

Qualifications

• Experience with capacity, demand, or supply planning for large-scale technical infrastructure.
• Build the forecasting model yourself using SQL against a data warehouse and Python-based tools.
• Understanding data-center resource classes and how storage/egress differ from GPUs.
• Ability to instrument forecast error and prefer transparent models over opaque ones.

Responsibilities

• Build and own the medium-range multi-resource demand forecast across resources (accelerators, CPU, storage, network, managed services).
• Run the plan-vs-reality loop and drive variance toward zero with planning-tools and data teams.
• Qualify each capacity tranche against forecast for shape, region, quarter and envelopes.
• Partner with Finance to turn forecasts into core drivers for spend and savings opportunities.

Skills

Capacity planning

Forecasting

SQL

Python

Pandas

Data interpretation

Education

Bachelor's degree

Tools

SQL

Python

Pandas

Job description

Anthropic is seeking an Infrastructure Capacity Planner to own medium-range demand forecasts for accelerators, CPU, storage, network, and managed services. You will interface with finance for cost efficiency and partner with data teams to refine planning tools and telemetry.

The role focuses on model-building using SQL and Python, with an emphasis on transparent forecasting and workload optimization. Based in San Francisco, some hybrid office time is expected.