Kevin Lan

• Harmonized multi-source secondary data (2-4 administrative claims and EHR feeds, ~2M patient records, multi-TB scale) into an OMOP CDM-style data model, mapping normalized (3NF) clinical fields to SNOMED CT, RxNorm, and LOINC; used AI-assisted development (Cursor, OpenCode) while owning spec-writing and output validation
• Created the metadata and data dictionary managing terminology and vocabulary mappings across the database, supporting consistent SNOMED CT, RxNorm, and LOINC usage system-wide
• Serve as primary client-facing data lead for the firm's largest (highest-revenue) pharma client, meeting weekly with clinical trial study designers, epidemiologists, clinical managers, and biostatisticians; client subsequently extended the RWD analysis contract
• Resolved a previously strained client relationship by introducing a standardized SAP template and client-facing data dictionary, since adopted company-wide; reduced error rates 50%+ and improved turnaround time 15%
• Review SAPs and RWD analyses authored by fellow team members (4 to date), ensuring methodological consistency and QA rigor prior to client delivery
• Designed and maintain Neo4j/Cypher graph models linking patient-drug relationships across ~2M patients; modeled 180,000+ IUI cycles linked to pregnancy and medication outcomes for observational IUI/IVF treatment-outcome research
• Authored 5 statistical analysis plans (SAPs), structuring multi-section deliverables (up to 7 sections per project), improving project completion rate 60%
• Established team-wide QA/validation standards via a SKILL.md repository for Cursor/OpenCode, adopted by other engineers, reducing errors 20% and cutting task completion time 40%
• Co-authored (3rd of 9 authors) "Real-World Insights on Controlled Ovarian Stimulation for IVF/ICSI: Data from a Large Fertility EMR Database in the US (2010-2024)," poster presentation, ISPOR 2026, Philadelphia, PA

• Configured and maintained Microsoft FHIR Server integrations, parsing FHIR resources into a relational structure across 65k+ patient records via an automated harmonization pipeline (MS Fabric, PySpark, OMOP CDM), cutting processing time 40%
• Served as data governance lead for a CDC-contracted project; created data models and metadata standards guiding engineering teams' product design decisions, and audited databases for validity and structural integrity
• Designed a data governance and validation workflow standardizing external source integration, cutting integration time from 3 months to 1 month
• Configured FHIR API bulk import of 10k patient records, improving import time 10-fold
• Built semantic metadata models spanning 4 infectious disease domains using GENEPIO ontologies, ingested into CDC's OAMD system
• Engineered a weekly-refreshed RAG pipeline (Azure ML) vectorizing 5GB of policy text for an AI agent, applying prompt engineering to reduce hallucination
• Created a GenAI RAG recommendation engine (cosine similarity, Azure AI Search) surfacing insights from regulatory/biomedical literature, awarded $10,000 from the Fast Prototyping Fund
• Built and demoed a Streamlit/GPT-4 recommendation bot for an HHS tech-innovation project proposal, presented to the leadership team overseeing the government contract

• Served as data architect for NIOSH's Adult Blood Lead Epidemiology Surveillance (ABLES) program, directing a contract engineer and owning the technical roadmap for migrating ABLES to Azure cloud infrastructure, cutting cloud costs 70% and reducing data refresh turnaround from 3 months to as fast as 1 week
• Led the OMOP CDM adoption initiative, harmonizing normalized clinical data to SNOMED CT, RxNorm, and LOINC, and led cross-division data-standard harmonization working sessions for DFSE
• Contributed to FHIR implementation guide development and maintained LOINC code sets, the PHIN-VAD value set, and the Message Mapping Guide
• Served as subject matter expert in CDC's OHDSI-Vulcan-OMOP work group, contributing to the International Patient Access Implementation Guide (IPA IG)
• Coordinated multi-state partner onboarding to the SAMS/SDX health information exchange platform
• Authored NORA Fund proposal integrating NIOSH auto-coding with EHR data, accepted, $30,000 granted
• Designed cloud data architecture (Azure Data Factory, Databricks Unity Catalog) and led FHIR API integration/USCDI+ research resulting in accepted submissions to USCDI v3 and USCDI+ v2, cutting state health department ETL/ingestion time from 6 months to 1 week
• Harmonized occupational health data across states with inconsistent units, fields, and coding systems to support ML model development for occupational health analytics

• Built statistical compartmental epidemiological models simulating global COVID-19 spread and trend trajectories, informing forecasting and response strategy
• Led manual data labeling (2M tweets, 3-person team) to build a misinformation gold-standard dataset; benchmarked Random Forest against BERT/RoBERTa NLP classifiers (best model: 89% F1) via Databricks fine-tuning, published in an internal report used by CDC leadership
• Applied BERTopic topic modeling and built a no/low-code R Shiny interface enabling CDC-wide users to query travel data without writing SQL, cutting policy review time from 3 months to 1 month
• Engineered a real-time ELT pipeline (Azure Data Factory, Synapse) ingesting 1TB+ per batch of travel and pandemic surveillance data every 3-7 days, feeding COVID-19 emergency response reports delivered to HHS and the White House emergency health response team
• Synthesized 5-disease forecasting and 15+ surveillance trend-analysis reports informing COVID, Monkeypox, and Ebola response
• Parsed and harmonized terabyte-scale outpatient antibiotics utilization data from HealthVerity's long-term care EHR database into a normalized (3NF) structure for COVID-era antibiotic stewardship reporting
• Published a white paper on USCDI and EHR data standardization and briefed CDC leadership, strengthening data elements accepted into USCDI v4
• Standardized travel data capture elements for USCDI+, coordinating with HHS ONC and CMS; submissions accepted in USCDI+ v1 and v2

• Designed the UI blueprint for a new EPIC Ambulatory Care patient admission page, adopted clinic-wide as the standard ambulatory care module
• Reviewed medication prescriptions and supported therapy management, improving vaccination rate 150% (2019–2020)
• Co-developed an IRB-approved observational study protocol on inhaled aminoglycoside antibiotic use in outpatient cystic fibrosis patients (~7,000 patient records); findings informed a published study and an internal practice guideline

• Standardized drug administration routes in the EPIC Willow pharmacy EHR environment and built Tableau/SlicerDicer dashboards, reducing non-standard routes 15% (INOVA Medical Center)
• Analyzed 5 years of Medicare Part D claims data to characterize biologics pricing trends and projected insurance-coverage impact (FDA)
• Built an automated NDC data-cleaning pipeline against RxNorm and First Databank drug data, and designed a drug-shortage data capture method for a medication diversion detection program (Kit Check)

2-year, doctoral-level appointment; travel-data epidemiology & emergency response focus

Doctor of Pharmacy (PharmD), Emphasis in Informatics & Research

BS, Biochemistry