AI Applied Scientist

Verily•San Bruno, CA

About The Position

Verily Health is a data platform and technology company purpose-built to power AI-enabled precision health solutions that accelerate research and improve care for individuals and communities. Uniquely positioned at the intersection of technology, data science, and healthcare, Verily transforms multimodal health data into insights, models, and actions that make healthcare more personalized, predictive, and precise. As an AI Applied Scientist, you will occupy a unique, highly impactful position that sits at the intersection of our real-world data (RWD) curation mission and our cutting-edge AI Agent development. In this role, you will be responsible for a balanced blend of designing scalable pipelines to curate Electronic Health Records (EHR) and claims data, and conducting advanced research into foundational healthcare models and intelligent multi-agent conversational architectures. You will apply frontier Large Language Models (LLMs) and advanced machine learning techniques to extract deep clinical meaning from messy, unstructured medical text, turning fragmented healthcare data into trustworthy, analysis-ready longitudinal datasets. Your work will directly ground our autonomous care agents and clinical decision support tools in robust, real-world clinical datasets, driving the next generation of healthcare transformation.

Requirements

Master’s degree in a quantitative discipline (e.g., Data Sciences, Computer Science, Biomedical Informatics, Statistics, Applied Mathematics, or equivalent practical experience).
Minimum of 3 years of industry experience applying advanced machine learning, NLP, and generative AI techniques (supervised/unsupervised learning, prompt engineering, agentic workflows) to clinical or healthcare datasets.
Direct experience working with and curating real-world data (such as EHRs or medical claims), with a deep understanding of the complexities and limitations of unstructured medical text.
Strong proficiency in Python and standard scientific computing/deep learning libraries (e.g., PyTorch, TensorFlow).

Nice To Haves

PhD degree in a quantitative discipline (e.g., Computer Science, Biomedical Informatics, Machine Learning, or related field).
Familiarity with advanced agent orchestration frameworks (e.g., LangGraph) and foundation model pre-training stacks (e.g., NVIDIA NeMo, Parabricks).
Familiarity with standard medical terminologies, vocabularies, and ontologies (e.g., SNOMED-CT, LOINC, RxNorm, ICD-10) and healthcare data models (FHIR, OMOP).
Experience working closely with clinical subject matter experts to establish ground truth benchmarks and adjudicate complex data abstraction guidelines.
Strong track record of scientific excellence, including peer-reviewed publications or submitted patents in healthcare AI/ML.

Responsibilities

Design, develop, and deploy advanced generative AI workflows and multi-agent conversational architectures (e.g., using LangGraph) to power personalized user interactions and automated symptom triage.
Implement, build on, and augment existing LLM/NLP tools to automate the abstraction of high-priority clinical variables and derived features from unstructured medical text, maximizing data completeness and accuracy.
Develop automated evaluation pipelines and "LLM-as-a-judge" grading rubrics to continuously benchmark agent safety, accuracy, factuality, and compliance across model updates. And enabling self-optimization cycles.
Handle real-world data challenges from clinical and remote settings, ensuring absolute safe data boundaries, privacy preservation, and rigorous technical validation before production releases.
Communicate highly technical results, methods, and evaluation frameworks clearly via presentations and well-structured reports to both technical and non-technical cross-functional audiences.