Research Engineer, Medical Imaging Data Quality & Evaluation

Epsilon Health; San Francisco (CA); On-site; USD 120,000 - 170,000; Full time

Job summary

Epsilon Health in San Francisco seeks a Research Engineer to own data quality and evaluation across our ML research efforts. This role focuses on ensuring clean, well-curated data for model training and robust evaluation of radiology report generation.

You will build filtering pipelines for VLM and classifier datasets, design metrics that capture accuracy, completeness, and reporting style, and collaborate with radiologists to set quality criteria and address edge cases.

Qualifications

• 2+ years of industry or research experience in ML, data engineering, or a related area.
• Strong Python and solid software engineering fundamentals; comfortable building tooling and data pipelines from scratch.
• Experience or interest in data quality, evaluation, or NLG/factuality assessment for generative models.

Responsibilities

• Build data filtering and curation pipelines to keep training sets clean and high quality at scale.
• Develop model-based data quality signals and active-learning loops for high-value cases.
• Collaborate with radiologists and annotators to define quality criteria and adjudicate edge cases.
• Design evaluation methodology for radiology report generation focusing on clinical accuracy, hallucination, and reporting style.
• Build and maintain clinical benchmark sets with stratified modality and pathology.

Skills

Python

Data pipelines

ML research

Analytical mindset

Tools

Spark

Airflow

Databricks

Job description

Epsilon Health in San Francisco seeks a Research Engineer to own data quality and evaluation across our ML research efforts. This role focuses on ensuring clean, well-curated data for model training and robust evaluation of radiology report generation.

You will build filtering pipelines for VLM and classifier datasets, design metrics that capture accuracy, completeness, and reporting style, and collaborate with radiologists to set quality criteria and address edge cases.