OCI AI Evaluation Science

This scientist will own evaluations from defining ambiguous business or customer questions through designing experiments, analyzing model behavior, and presenting conclusions to leadership. The work is part of OCI’s AI Evaluation Science function and spans areas including reasoning, coding, retrieval-augmented generation, agents, multimodal systems, multilingual performance, and responsible AI. The role involves building reproducible pipelines, automated evaluators, human-review programs, and reusable benchmarks while assessing quality alongside safety, reliability, cost, latency, and data integrity. The successful candidate will collaborate across research, engineering, product, data, and operations teams and may publish new evaluation methods or convert them into production capabilities. It suits an independent, technically strong applied scientist who combines machine learning expertise, software engineering discipline, statistical judgment, and clear stakeholder communication.