Machine Learning Engineer, Evaluation

About Bask Health

Bask Health is a software platform that enables entrepreneurs, doctors, and healthcare businesses to launch, manage, and grow direct-to-consumer telehealth businesses. Our platform brings together patient experiences, clinical workflows, pharmacy operations, and commerce tools in one place. We're an entrepreneurial, product-focused engineering team that values ownership, technical excellence, and building software that solves real problems. Engineers work closely with product teams and company leadership, take features from idea to production, and have a direct role in shaping the platform. Engineering at Bask is AI-first and builder-led. We use tools like Cursor and Claude Code throughout development while maintaining high standards for reliability, testing, and software quality.

About the Role

We're looking for a Machine Learning Engineer, Evaluation & Reliability to help ensure that AI-powered features across Bask Health are accurate, dependable, and ready for real-world use. Bask Health uses software to connect patient experiences, merchant operations, clinical workflows, pharmacy coordination, and commerce capabilities. As AI becomes more integrated into these systems, we need reliable ways to measure model performance, identify failures, and verify that changes improve the user experience without introducing new risks. You'll help develop the evaluation infrastructure, testing workflows, and performance measurements used to assess large language models, retrieval systems, AI assistants, and automated workflows. This is a hands-on engineering role that combines machine learning evaluation, software development, data analysis, and production quality assurance. You'll work with model outputs, evaluation datasets, automated testing tools, and application telemetry to understand how AI systems perform under realistic conditions. Your work may include creating evaluation datasets, implementing automated benchmarks, measuring response quality, testing retrieval accuracy, analyzing model failures, and monitoring production behavior. You'll collaborate with applied AI engineers, machine learning engineers, backend engineers, product managers, and other stakeholders to define what successful AI behavior looks like for different workflows. You'll help translate those expectations into measurable tests and actionable engineering improvements. Because Bask operates in healthcare-related environments, AI evaluation must account for accuracy, privacy, appropriate permissions, and the consequences of unreliable responses. You'll help identify situations where AI systems produce unsupported information, expose restricted data, attempt unauthorized actions, or operate outside approved clinical and operational boundaries. This position emphasizes practical implementation and continuous improvement. You'll contribute to evaluation methodologies and engineering standards while working with senior engineers on broader reliability strategies and more complex technical decisions. We're looking for an engineer who enjoys investigating how AI systems behave, can turn ambiguous quality problems into measurable tests, and understands that reliable AI products require ongoing evaluation—not just successful demonstrations.

What You'll Do

Build and maintain evaluation systems for AI-powered applications across Bask Health. Develop automated testing frameworks for large language models and AI application workflows. Create and maintain representative datasets for evaluating model behavior. Implement benchmarks that measure response accuracy, relevance, completeness, and consistency. Develop evaluation pipelines for prompts, retrieval systems, and model configurations. Test AI assistants and agents against realistic user scenarios and workflow requirements. Build automated regression tests to detect performance changes between releases. Measure hallucination rates, unsupported claims, and other model quality failures. Evaluate retrieval-augmented generation systems for document relevance and answer grounding. Develop tests for structured model outputs and schema compliance. Evaluate tool-calling accuracy and agent action selection. Test AI workflows for unauthorized actions, incorrect permissions, and unsafe behavior. Implement evaluation checks for sensitive patient and operational information. Build dashboards and reports that communicate AI quality metrics to engineering teams. Analyze evaluation results and identify recurring failure patterns. Investigate production incidents involving inaccurate or unreliable AI outputs. Collaborate with applied AI engineers to improve prompts, retrieval logic, and application behavior. Support model comparisons and controlled experiments. Evaluate tradeoffs between model quality, latency, reliability, and inference cost. Develop automated quality checks for CI/CD and AI feature release workflows. Implement monit