Design, build and operate a substantial part of the autonomy evaluation metrics platform, collaborating with Perception, Planning and Systems engineers to connect model experiments to reliable metrics and improve models and data.
Compensation
Salary Range for California Based Applicants: $179,000-$223,800
Responsibilities
Own the design and delivery of a metrics-platform area, such as the metric registry, computation services or comparison and reporting APIs
Design versioned metric definitions and data models for inputs, aggregation, slicing and results
Build distributed computation and storage paths with idempotent execution, incremental processing and controlled backfills
Build comparison tools that expose coverage, sample sizes, uncertainty and changes in evaluation inputs
Partner with model developers to design evaluation experiments, inspect predictions and analyze failure patterns
Rivian’s Autonomy Eval Infrastructure team builds the software that helps engineers understand and improve the ML models behind our driving systems and gives release teams evidence they can trust. As a senior metrics-platform engineer, you will design, build and operate a substantial part of the platform that makes autonomy evaluation consistent and reproducible.
Locations (3)
Palo Alto, California, United States
London, England, United Kingdom
Belgrade, Central Serbia, Serbia
Compensation
Salary Range for California Based Applicants: $179,000-$223,800 (actual compensation will be determined based on experience, location, and other factors permitted by law).
Responsibilities
Own the design and delivery of a metrics-platform area, such as the metric registry, computation services or comparison and reporting APIs
Design versioned metric definitions and data models for inputs, aggregation, slicing and results
Build distributed computation and storage paths with idempotent execution, incremental processing and controlled backfills
Build comparison tools that expose coverage, sample sizes, uncertainty and changes in evaluation inputs
Partner with model developers to design evaluation experiments, inspect predictions and analyze failure patterns
