Data Scientist

Job description

Many structural heart patients suffer from heart failure with limited options. Our Implantable Heart Failure Management (IHFM) team, part of the AI, Product and Platforms organization, is at the forefront of addressing these unmet patient needs through pioneering technology that enables early, targeted therapeutic intervention. Our innovative solutions are not just transforming patient care but also creating a unique and exciting environment for our team members. It is our driving force to help patients live longer and healthier lives. Join us and be part of our inspiring journey. At Edwards Lifesciences, the Implantable Heart Failure Management (IHFM) AI, Product and Platforms organization designs and builds the software and data products that clinicians and patients depend on. As a Senior Data Scientist, you own modeling for a clinical or product domain, drive problem framing with product and clinical partners, own offline-validation rigor, and author model documentation that meets FDA submission expectations for regulated products. Based in Irvine, CA, you'll join a high-impact medtech innovation hub in the heart of Orange County, collaborating in person with cross-functional teams to shape patient-focused technology. How you'll make an impact Domain modeling. Own modeling for a clinical or product domain, for example, arrhythmia classification, cardiac imaging segmentation, clinical prediction, or patient outcome forecasting, from framing through offline validation. Architecture selection. Select and adapt architectures for the modality (vision transformers or encoder-decoder networks for imaging, transformers, or temporal models for signals) and apply self-supervised and representation learning under limited labels. Statistical depth. Apply survival analysis, Bayesian modeling, and causal inference where the clinical question requires (statsmodels, lifelines or scikit-survival, PyMC). Evaluation & subgroup. Design offline evaluation, calibration, and clinical performance validation strategies, including subgroup and bias analysis across sex, age, and device cohort, with Medical Affairs and Clinical Science. Interpretability & documentation. Author submission-grade model documentation including calibration, uncertainty, and interpretability evidence (SHAP, Captum), and improve the team's modeling and validation practice. Research adoption. Evaluate and adapt current AI/ML for health research to IHFM problems, and mentor less experienced data scientists. Handoff. Coordinate the handoff of offline-validated custom models to AI/ML Engineers (Applied), and partner with Regulatory Affairs on submission strategy. Algorithm development, analysis & reporting. Create, test, and improve complex algorithms, NLP, and machine learning models; analyze results; develop insights; and produce reports and dashboards to communicate performance and findings to stakeholders. Data preparation & quality. Process, cleanse, label, and verify structured and unstructured data used for analysis; collaborate with internal teams and external partners on data standards, metrics, analytics, and reporting. Tools, integration & design controls. Identify and integrate data sources, software, and analytics tools (e.g., Python, SQL, SAS, Power BI, Tableau Prep); support development of design control documentation including algorithm requirements and risk documentation. What you'll need (Required): Bachelor's in related field (e.g., Computer Science, Engineering, Biostatistics or Scientific) plus 2 years of previous experience including industry or industry/ education This is an onsite role based in Irvine, CA. Relocation assistance is not provided, and candidates must reside within a 50-mile radius of Irvine to be considered. What else we look for (Preferred): Depth in at least one data modality relevant to IHFM, for example, time-series, or medical imaging (echocardiography, cardiac magnetic resonance imaging, or computed tomography). Command of the relevant architecture families and of transfer and self-supervised learning. Strong Python and SQL, and depth in modeling and experiment tracking such as PyTorch, TensorFlow, and MLflow. Command of evaluation, calibration, and clinical performance methodology, including subgroup analysis, and the ability to author documentation suitable for regulatory submission. A track record of owning modeling for a domain and communicating results to clinical, product, and regulatory partners. Experience with clinical validation, retrospective clinical data, or regulated medical software (SaMD). Depth in time-series or medical imaging modeling, including segmentation and registration, for example, MONAI, pydicom, or SimpleITK. Bayesian modeling, causal inference (DoWhy, EconML), or survival analysis (lifelines, scikit-survival). Generative approaches for augmentation or synthetic data (autoencoders, GANs, or diffusion models). Exposure to multimodal modeling, combining signals, imaging, labs, and notes into patient-state models. Familiarity with R, JAX, or hyperparameter optimization (Optuna, Ray Tune). Contributions to internal or external research (publications, patents, or venues such as MICCAI or ML4H). Aligning our overall business objectives with performance, we offer competitive salaries, performance-based incentives, and a wide variety of benefits programs to address the diverse individual needs of our employees and their families. For California (CA), the base pay range for this position is $87,000 to $123,000 (highly experienced). The pay for the successful candidate will depend on various factors (e.g., qualifications, education, prior experience). Applications will be accepted while this position is posted on our Careers website. Edwards is an Equal Opportunity/Affirmative Action employer including protected Veterans and individuals with disabilities.