Senior AI Data Scientist

About the Role

As a Senior AI Data Scientist at Beth Israel Lahey Health, you will serve as a technical leader driving complex machine learning and generative AI initiatives. You are responsible for designing, fine-tuning, and evaluating advanced AI models specifically tailored to healthcare environments.

Key Responsibilities

• Lead the end-to-end design, development, and evaluation of cutting-edge ML and Generative AI solutions across diverse clinical use cases.
• Act as a subject matter expert on Large Language Models (LLMs), providing deep insights into architecture tradeoffs, inference behaviors, strengths, limitations, and safety constraints.
• Architect and oversee LLM fine-tuning strategies, including instruction tuning and parameter-efficient methods, ensuring strict alignment with domain needs and governance standards.
• Define robust model evaluation frameworks covering quantitative metrics, qualitative reviews, bias analysis, hallucination assessment, and rigorous safety testing.
• Optimize prompt engineering strategies to ensure generative AI outputs are robust, consistent, interpretable, and clinically relevant.
• Pioneer Retrieval Augmented Generation (RAG) solutions, focusing on high-quality data selection, grounding accuracy, relevance evaluation, and performance analysis.
• Conduct advanced exploratory data analysis (EDA), feature engineering, and statistical modeling to support predictive and prescriptive analytics for healthcare outcomes.
• Translate complex clinical challenges into clear, actionable modeling strategies and comprehensive evaluation plans.
• Mentor entry-level and mid-level data scientists through rigorous model reviews, best practice guidance, and collaborative problem-solving sessions.
• Partner cross-functionally with data engineering, software development, and clinical teams to ensure AI models are seamlessly deployable, continuously monitored, and fully compliant with regulatory requirements.
• Champion Responsible AI principles, fostering a culture of fairness, bias mitigation, transparency, thorough documentation, and appropriate clinical application.
• Communicate intricate model behaviors, tradeoffs, and outcomes effectively to technical leaders, clinicians, and executive stakeholders.