Hire Lambert Leong, PhD | Health AI Applied Scientist

Health AI Applied Scientist

Lambert Leong, PhD

Healthcare is moving from isolated machine learning models toward AI systems that need to be evaluated, monitored, and trusted in real clinical workflows. My work sits in that gap: I build and evaluate Health Artificial Intelligence (Health AI) systems across clinical Generative Artificial Intelligence (GenAI), healthcare foundation models, recommender systems, structured Electronic Health Record (EHR) modeling, Current Procedural Terminology (CPT) sequence modeling, and medical imaging. In other words, I focus on turning advanced AI methods into measurable healthcare tools that can stand up to clinical, product, and governance scrutiny.

Email Lambert View Resume

Best Fit

I am most interested in senior applied scientist and Health AI roles where the work combines ambiguous product discovery, rigorous model evaluation, and real healthcare impact. I am especially drawn to teams building clinical GenAI evaluation and safety systems, healthcare foundation models, structured EHR or CPT models, recommender systems for care navigation, and multimodal medical imaging products.

Selected Proof Points

  • Led clinical GenAI evaluation work across five healthcare use cases by combining automated metrics, Large Language Model (LLM)-as-judge methods, adversarial prompting, calibrated clinical Subject Matter Expert (SME) review, reusable evaluation libraries, and monthly monitoring dashboards.
  • Supported governed review for clinical AI systems spanning patient message response, chart summarization, medical coding, expanded response generation, and Electronic Medical Record (EMR) text transformation, with $1.5M realized business benefit and $5M projected annual benefit reported by the business.
  • Built SparseEHR, a scalable structured EHR foundation modeling approach accepted to the SD4H Workshop at the International Conference on Machine Learning (ICML) 2026, using a LLaMA-style autoregressive model and custom CPT tokenizer trained on service patterns from 35M members.
  • Recognized with a 2026 UnitedHealth Group / Optum Make IT Happen Award for healthcare foundation-model work connected to SparseEHR.
  • Designed a world-model-inspired healthcare research direction over 500M service timepoints across 35M members, enriched with date-of-service and demographic signals to study future service-pattern prediction.
  • Published and contributed to medical imaging work across Computed Tomography (CT) segmentation, spleen volumetry, body-composition analysis, breast cancer risk prediction, generative medical imaging, and self-supervised whole-body X-ray embeddings.

Related Work to Explore

For recruiters and hiring teams who want more context, these internal write-ups connect the summary above to specific projects, publications, and technical perspectives.

Agentic AUC and Model Evaluation

A practical discussion of how agentic systems can be evaluated with area under the receiver operating characteristic curve (AUC) when outputs are not naturally continuous.

Deep Learning for Breast Cancer Risk

Mammography-based risk prediction work focused on interval and screen-detected cancer, including clinical evaluation against established risk factors.

CT, Body Composition, and Segmentation

Postdoctoral work on computed tomography (CT) imaging, body-composition pipelines, natural language processing case finding, and spleen segmentation.

Where I Add Value

Clinical GenAI Evaluation and Safety

I design evaluation plans that connect automated metrics with clinical SME review, adversarial testing, and continuous monitoring. However, I do not treat evaluation as a dashboard exercise; the goal is to understand whether a model behaves reliably enough for the clinical workflow it is entering.

Healthcare Foundation Models

I build large-scale sequence models for structured healthcare data, including EHR and CPT modeling. In other words, I work on the less flashy but high-impact side of foundation models: learning from longitudinal healthcare events and connecting those representations to useful recommendation and planning problems.

Medical Imaging and Clinical Evidence

My PhD and postdoctoral work focused on medical imaging, cancer risk prediction, quantitative biomarkers, and CT analysis. That background helps me evaluate AI work through both a technical lens and a clinical evidence lens.

Cross-Functional Product Leadership

I have led ambiguous discovery and testing work across clinical, product, engineering, data science, and business teams. As such, I am comfortable translating between model behavior, clinical risk, implementation constraints, and executive decision-making.

Technical Focus

  • Clinical GenAI
  • AI governance
  • LLM evaluation
  • Foundation models
  • Structured EHR
  • CPT tokenization
  • Recommender systems
  • Longitudinal modeling
  • Medical imaging
  • Natural Language Processing (NLP)
  • Computer vision
  • PyTorch

Final Thoughts

I care about Health AI that is technically strong, clinically grounded, and measured honestly. If your team is building AI systems that need to work in real healthcare settings, I would be glad to connect at leonglambert@gmail.com or through my resume.