I lead clinical data science and statistical programming teams, and I help organizations move to open-source R and Python analytics that regulators can trust: reproducible pipelines, CDISC-native data, and AI that runs inside the GxP boundary rather than around it.

• Associate Director, Statistical Programming @ Kardigan. I manage a programming team, lead statistical programming for the tonlamarsen cardiovascular program, and architected (with IT and QA) a governed clinical analytics platform, from validation framework to user training and adoption.
• At Merck, I helped drive the move to R. I architected an enterprise GenAI (LLM/RAG) platform whose SAS-to-R conversion tooling was adopted organization-wide.
• At Merck, I also led statistical programming for LIPFENDRA® (enlicitide) from Phase 2b through the pivotal Phase 3 CORALreef program and owned the NDA eSubmission packages (Define-XML, eCTD, ADRG). FDA approval: July 2026.
• PhD in Data Science (2026), with research on AI/ML governance, validation and automation in clinical development.
• Currently exploring digital protocols with CDISC USDM and ICH M11 (usdm-toolkit).

flowchart TB
    P["Protocol & SAP<br/>USDM · ICH M11"] --> S["SDTM"]
    S --> A["ADaM<br/>admiral · admiralagent"]
    A --> T["TLFs<br/>tlfstack · py4csr"]
    T --> E["eSubmission<br/>Define-XML · eCTD · ADRG"]
    AI(["🤖 LLM agents + validation gates"]) -.-> A
    AI -.-> T
    G(["🛡️ GxP: validation · audit trail · Part 11"]) -.-> S
    G -.-> E
    L(["👥 People · SOPs · CRO oversight · training"]) -.-> S
    L -.-> T
    classDef core fill:#0f766e,stroke:#2dd4bf,color:#ffffff
    classDef layer fill:#4c1d95,stroke:#a78bfa,color:#ffffff
    class P,S,A,T,E core
    class AI,G,L layer

• Workflow topology–guided skill loading for agentic AI in clinical trial data pipelines: sequential exploratory evaluation. JMIR Formative Research, 2026. doi:10.2196/100769
• A non-destructive methodological framework for modernizing legacy clinical reporting systems for AI-driven pharmacoinformatics: a SAS case study. Frontiers in Pharmacology, 2026. doi:10.3389/fphar.2026.1876207 · code
• CAVE-Onc: graph-constrained agentic validation for cross-domain contradictions in CDISC oncology submissions. PLOS ONE, 2026. doi:10.1371/journal.pone.0350376 · code
• ClinAgent: AI-assisted methodology for clinical trial data processing and statistical programming. Biology Methods and Protocols (Oxford University Press), 2026. doi:10.1093/biomethods/bpag032

_{Plus 13 PhUSE / PharmaSUG conference papers (2023–2026). Full list on Google Scholar.}

• AI in Validated Environments: LLMs Inside the GxP Boundary _{· Nov 04, 2026}
• Visit Windowing, Baseline, and LOCF: The Rules That Decide BDS _{· Nov 04, 2026}
• Define-XML and the Reviewer's Guide, Explained for Programmers _{· Oct 26, 2026}
• How to Read a Protocol and SAP as a Statistical Programmer _{· Oct 26, 2026}
• Git for SAS Programmers: Version Control in a GxP World _{· Oct 06, 2026}

_{From jaimeyan.com: practical articles on CDISC, R in pharma, GxP and LLM automation. RSS}

_{All demos and benchmarks use synthetic or public test data (e.g. the CDISC Pilot). Views are my own and not those of any employer.}