Curriculum Vitae, Narrative, & Portfolio

William T. Farrington, M.D.

Acute Care Hospitalist — ABIM-Certified, PA-Licensed Data Scientist & Engineer

01

Supporting Documents

Curriculum Vitae PDF · August 2026
Professional Narrative PDF · August 2026
02

Professional Narrative

I am a board-certified internist and acute care hospitalist driven by a well-known engineering maxim: “all models are wrong, but some are useful.” After engineering and medical school at Texas A&M and internal medicine residency in Dallas, I spent two years on the McGovern Medical School faculty, practicing at Memorial Hermann Hospital–Texas Medical Center. I concluded that appointment in the spring of 2026 to relocate to Philadelphia with my fiancée; now licensed in Pennsylvania, I continue public and private data work while seeking a clinical home in which to combine inpatient care, teaching, and responsible technical work.

My foundation is high-volume, high-acuity inpatient medicine in a hospitalist group responsible for more than 40% of the institutional admissions, on service lines spanning oncology, neurology, transplant, geriatrics, and trauma. My group’s triage and capacity-management duties entailed ED- and ICU-facing acuity forecasting, level-of-care decision-making, and transfer requests from outside institutions. My teaching roles included: supervising residents and delivering triage lectures for incoming interns, disease-management lectures for surgical residents, and medication reconciliation lectures during safe-discharge workshops.

Pilot deployments of Viz.ai, Ambiance, and Regard at my prior institution left me unwilling to dismiss AI tools or accept it casually. These tools remain under-acknowledged risk surfaces, so I strictly define risk by what software may read and what its output may induce. The projects listed on this page are, to date, built and maintained myself in Python, SQL, and Bash. They hold to that same standard: only mimicwarehouse faces de-identified patient data, kept local and redundantly inaccessible to agents, while other clinically adjacent prototypes run on synthetic or fictional inputs. Attending physicians must adopt these tools early and earnestly enough to catalog their hazards and to lobby for privacy and quality before profit. Ultimately, I believe the core essence of modern medicine remains at the bedside.

03

Curriculum Vitae

Positions & appointments

  1. 04/2024 – 03/2026

    Hospitalist, Acute Care (Internal) Medicine

    Memorial Hermann Hospital–Texas Medical Center, Houston, TX

  2. 04/2024 – 03/2026

    Assistant Professor of Emergency Medicine, Hospitalist Division

    McGovern Medical School at UTHealth, Houston, TX

Education & training

  1. 06/2020 – 06/2023

    Internal Medicine Internship & Residency

    Texas Health Presbyterian Hospital, Dallas, TX

  2. 07/2015 – 05/2020

    Doctor of Medicine

    Texas A&M College of Medicine, College Station, TX

  3. 08/2011 – 05/2015

    Bachelor of Science in Biomedical Engineering, cum laude

    Texas A&M University, College Station, TX

Informatics & quality service

  1. 01/2026 – 03/2026

    Pilot User & Edge-Case Tester, Hospitalist Cohort

    Regard AI EHR Copilot Implementation Team

  2. 08/2025 – 12/2025

    Clinical Beta Tester, Hospitalist Cohort

    Ambiance AI Scribe Beta Utilization Team

  3. 04/2025 – 06/2025

    Epic SlicerDicer & Inter-Operability Analyst

    Capacity Management Dashboard Analytics Task Force

  4. 08/2024 – 04/2025

    Viability Analyst & Hospitalist Cohort Representative

    Viz.AI Aortic Dissection Classifier Implementation Team

  5. 08/2024 – 03/2026

    Active Member & Root-Cause Analyst

    Hospitalist Division Quality Assurance Committee

At a glance

  • Certification American Board of Internal Medicine, 10/2023 – present
  • Licensure Pennsylvania, 05/2026 – present / Texas, 11/2023 – present
  • NPI 1154940732
  • Clinical focus Triage & inpatient capacity management; EHR heuristics & informatics — risk stratification, medication reconciliation, continuity optimization
  • Teaching Intern lecture series, McGovern Medical School; resident workshops, Memorial Hermann Hospital–Texas Medical Center
  • Language Spanish — working clinical proficiency
  • Selected work Peer-reviewed case report, two abstracts, an open-access WikEM entry, and an ATS poster — full list in the CV
04

Active Projects

Public

hupsim

V1 (demo) shipped

A room-level capacity model of a single academic hospital campus, built entirely from cited public sources with a confidence tag on every structural fact and all dynamics explicitly simulated.

  • simulation
  • capacity modeling
  • geospatial
  • provenance
hupsim — 3D campus view of the Hospital of the University of Pennsylvania with per-building census counts and the level-of-care hospital dashboard below
3D campus view & hospital dashboard
hupsim — room-level unit view of a general medicine floor with per-room occupants, nurse assignments, and a patient inspector panel
Room-level unit view & patient inspector

mimicwarehouse

V1 in progress

A local EMR data warehouse over MIMIC-IV with end-to-end provenance, protocol-frozen inquiry over retrospective data, and a pre-commit leak guard — data never leaves the machine.

57 of 172 session briefs · as of September 2026

  • Next Measurement-process summaries, then the cohort engine — specs, compiler, and attrition diagrams.
  • Then Event-aligned timelines and the protocol-freeze registry, closing the concepts & quality phase before the lab app’s first pages.
  • MIMIC-IV
  • DuckDB
  • Polars
  • Streamlit
  • data governance

medrecsim

V1 in progress

A medication reconciliation simulator for students, residents, and fellows — a resident-facing example of what physician-driven software design can produce, on wholly synthetic inputs.

12 of 39 session briefs · as of September 2026

  • Next Scoring, signature and debrief data behind a golden-test harness, plus the first formulary wave.
  • Then The accessible app shell and one complete reviewed case, playable end to end — the v0.1 slice.
  • simulation
  • medication reconciliation
  • medical education
  • synthetic data

phillysim

V1 in progress

Measuring census-tract access to health-relevant community resources across Philadelphia, with uncertainty shown beside every result — it measures access; it does not model outcomes.

18 of ≈ 40 work packets · as of September 2026

  • Next A fresh-clone reproducibility checkpoint covering the travel-time runs just completed.
  • Then Full ingest — every source snapshotted, duplicate locations reconciled, opening hours parsed.
  • civic data
  • public data
  • geospatial
  • transit access

Private

epppsynth

Private

A structured synthesis of existential psychotherapy literature for physicians and patients — corpus work applied to the parts of clinical life that resist quantification.

  • corpus synthesis
  • physician wellbeing

wikemgapsim

Private

Modeling editorial coverage gaps in WikEM, the open emergency medicine reference — where the clinical knowledge commons is thin, and whom that thinness reaches.

  • knowledge commons
  • text analysis
  • gap analysis

chesscomhistory

Private

A decade of personal game records as a longitudinal backbone for cross-analysis — a controlled sandbox for time-series and behavioral methods before they meet clinical data.

  • time series
  • API ingestion
  • methods sandbox

Some private work involves data that cannot be redistributed; the rest is simply unfinished. Neither is listed here as a placeholder for something that does not exist — ask, and I will walk you through any of it.

05

Technical Toolkit

Languages

Python, SQL, Bash

Data engineering

DuckDB, Parquet, Polars, pandas, reproducible pipelines, provenance & disclosure discipline

Clinical data & EHR

MIMIC-IV, Epic SlicerDicer, ICD-10 terminology, de-identification, risk stratification & medication reconciliation heuristics

Modeling & delivery

Discrete-event & capacity simulation, geospatial analysis, Streamlit, Git-based version control