# James Weatherhead

**MD/PhD/MPH candidate · AI Engineer · UTMB**

I build AI for medicine — and study how to run large language models on real patient data without leaking a patient's identity.

## At a glance
- **Publications:** 3 peer-reviewed (2026)
- **Recognition:** Winner, Google DeepMind Gemini 3 Pro hackathon — top 50 of 4,096 submissions
- **Deployed:** AutoTox — 66% faster sign-out, 10× review throughput at UTMB
- **Featured in:** UTMB News · Microsoft HLS Blog · Retool

## Focus
My doctoral work — advised by Peter McCaffrey, MD (Chief AI Officer, UTMB) — sits on a hard boundary in medicine. Healthcare AI can run under HIPAA Business Associate Agreements, but most genuinely useful tools (web search, current-literature retrieval) are not covered, so any query carrying protected health information must be reliably de-identified *before it leaves the hospital* — otherwise clinicians fall back on unsanctioned "shadow AI." My dissertation attacks that boundary with three contributions:

- **ASQ-PHI** — an adversarial synthetic-data benchmark for clinical PHI de-identification (rule-based, cloud, vector, and on-device/Ollama approaches). Peer-reviewed (Data in Brief); AMIA 2026 Amplify poster.
- **K-anonymity decay** — quantifying how re-identifiability erodes across multi-turn clinical LLM conversations. Peer-reviewed (Frontiers in Digital Health).
- **Local PHI Scrubber** — a Chrome extension doing on-device, human-in-the-loop PHI redaction with small (3–4B) models, so zero data leaves the machine while external tools stay usable. AMIA 2026 Amplify oral, Session CI25.

Before the MD/PhD I studied biology at St. Mary's University (minor in chemistry) and played NCAA tennis; born in London, I moved to South Africa at 13 (Pretoria Boys High School) before coming to the United States.

## Builds
- **VibeRad — Reimagining Radiology Education with Gemini 3 Pro** *(★ Winner, Google DeepMind Gemini 3 Pro hackathon — top 50 of 4,096)*: an AI teaching assistant that sees what you see — point at an MRI, ask, and it explains at your level, high-school to resident.
  - Write-up: https://james.jcweatherhead.com/viberad/
  - Live demo: https://aistudio.google.com/apps/drive/1pCMrggBue_gNE9IhzjN-KEdLm8cVPGjW?fullscreenApplet=true
  - Source: https://github.com/JamesWeatherhead/VibeRad
- **AutoTox** (deployed, UTMB): AI-powered toxicology interpretation live in clinical pathology — 66% faster sign-out (15→5 min), 10× throughput (50→500 cases/week), built in 5 weeks on a self-hosted Retool + private Azure (HIPAA) stack, with a pathologist verifying every report.
  - Write-up: https://james.jcweatherhead.com/autotox/
  - Case study: https://retool.com/blog/how-utmb-improved-diagnostics-ai
- **Local PHI Scrubber** (research): on-device, human-in-the-loop PHI redaction Chrome extension using small (3–4B) models — zero data transmission, while external tools stay usable. The applied third leg of the dissertation (alongside ASQ-PHI and K-anonymity decay). AMIA 2026 Amplify oral, Session CI25.
  - Write-up: https://james.jcweatherhead.com/local-phi-scrubber/
- **Imaging Cohorts by Natural Language** (UTMB × Microsoft Research): semantic search over medical-image embeddings and radiology-report indexing on Azure — collapsing weeks of PACS and data-warehouse coordination into a plain-language query. Selected contributor; named in the acknowledgements of Microsoft's HLS write-up.
  - Write-up: https://james.jcweatherhead.com/imaging-cohorts/
  - Microsoft HLS blog: https://techcommunity.microsoft.com/blog/HealthcareAndLifeSciencesBlog/using-natural-language-to-build-healthcare-imaging-cohorts-for-research/4472603
- **Biological-age agents** (Thrive AI Health, SF): biological-age models from wearable and biomarker data + LangGraph agents recommending personalized interventions (Junction data integration, LangSmith tracing); shipped to iOS and Android.
  - Write-up: https://james.jcweatherhead.com/thrive/

## Research (peer-reviewed, 2026)
1. Weatherhead J, Hasan A, Weatherhead J, Golovko G, Grant B, Garcia JD, Certuche HS, Powell RP, Abril JM, McCaffrey P. **K-anonymity decay in multi-turn clinical large language model conversations.** *Frontiers in Digital Health.* 2026;8:1832168. https://doi.org/10.3389/fdgth.2026.1832168
2. Weatherhead J, Golovko G, McCaffrey P. **ASQ-PHI: An adversarial synthetic data benchmark for clinical de-identification and search utility.** *Data in Brief.* 2026;65:112586. https://doi.org/10.1016/j.dib.2026.112586
3. Almomani M, Valaparla V, Weatherhead J, Fang X, Dabi A, Li C-Y, McCaffrey P, Hier D, Rodríguez-Fernández JM. **Evaluation of multiple generative large language models on neurology board-style questions.** *Frontiers in Digital Health.* 2026;7:1737882. https://doi.org/10.3389/fdgth.2025.1737882

Full publications — plain-language summaries, DOIs, and BibTeX: https://james.jcweatherhead.com/research/ (Markdown: https://james.jcweatherhead.com/research/index.md)

## Talks
- **AI in Health Conference (AIHC), Rice University — Sep 2026 (poster):** Agentic real-world-evidence analysis: EBV-specific neurovascular risk after corticosteroid exposure.
- **AMIA 2026 Amplify, Denver — May 2026 (oral, Session CI25):** Local PHI Scrubber: On-Device PHI Redaction in Clinical LLM Workflows.
- **AMIA 2026 Amplify, Denver — May 2026 (poster):** ASQ-PHI: An Adversarial Synthetic Benchmark for Clinical PHI De-Identification (sponsored by CU Anschutz Center for Health AI).
- **Digital Pathology Association Conference — Oct 2025:** Beyond Digital Pathology: AI to Improve Cancer Detection and LLMs to Personalize Care.
- **Retool Summit — 2024:** How UTMB Improved Diagnostic Testing for Thousands of Patients Using AI.
- **Perplexity AI HQ, San Francisco — 2024 (invited):** HIPAA-compliant semantic search for clinical content discovery.

## Press
- **UTMB News — Jun 2026:** UTMB AI Innovators Win International Hackathon with Radiology Viewer and Teaching Tool. https://www.utmb.edu/news/article/utmb-news/2026/06/26/utmb-ai-innovators-win-international-hackathon-with-radiology-viewer-and-teaching-tool
- **Microsoft Healthcare & Life Sciences Blog — Dec 2025:** Using Natural Language to Build Healthcare Imaging Cohorts for Research. https://techcommunity.microsoft.com/blog/HealthcareAndLifeSciencesBlog/using-natural-language-to-build-healthcare-imaging-cohorts-for-research/4472603
- **Retool Engineering Blog — Sep 2024:** How UTMB Improved Diagnostic Testing for Thousands of Patients Using AI. https://retool.com/blog/how-utmb-improved-diagnostics-ai

## Experience
- **AI Engineer — University of Texas Medical Branch** (2024–2026, Galveston, hybrid): with Peter McCaffrey, MD; built and deployed AutoTox; HIPAA-compliant semantic search and AI-agent infrastructure; the UTMB × Microsoft Research imaging-cohort collaboration.
- **AI Researcher — Thrive AI Health** (2025, San Francisco; OpenAI Startup Fund–backed): biological-age models from wearables; LangGraph agents; shipped to iOS/Android; debugging with LangSmith.
- **Research Assistant — Stephen Molldrem, PhD** (UTMB): CHIRP digital-health platform; CDC Social Vulnerability Index metrics.
- **Research Assistant — Amina El Ayadi, PhD** (UTMB): lipid biomarkers in neuroinflammation, murine burn models.
- **Research Assistant — Terry Shackleford, PhD** (St. Mary's University): anti-cancer drug testing on pediatric sarcoma cell lines.
- **Research Intern — ELM Lab, UT Dell Medical School:** meta-analysis of memory enhancement via deep brain stimulation in treatment-resistant epilepsy.

## Teaching & service
- **Guest Lecturer — "Agentic Coding with Claude Code"** (NTD 6088, UTMB Graduate School of Biomedical Sciences, Spring 2026): agentic coding workflows and tooling patterns with the Claude Code CLI, for a room of technical programmers and a PhD researcher.
- **Co-President & Co-Founder — AI in Medicine Student Organization, UTMB** (Aug 2022–Dec 2024): founded and led UTMB's student org for AI in healthcare — bi-weekly meetings on emerging clinical AI, K–12 coding camps, and educational talks for the UTMB community.
- **Medical-student leadership** (2022–2024): TMA Delegate & Leadership Committee Co-Chair; member, AMA Committee on Health Information Technology; UTMB School of Medicine Student Ambassador.
- **Clinical & community volunteering:** St. Vincent's Hope Clinic, Luke Society Clinic, and Street Medicine; earlier co-founded a STEM outreach RSO at St. Mary's and ran COVID-19 community outreach with the San Antonio Food Bank.

## Memberships & competitions
- **OpenAI Forum — Member** (Jun 2026–present): OpenAI's curated community advancing public discourse on safe AI development among researchers, developers, and stakeholders.
- **Kaggle — Competitions participant** (Jun 2024–present): NVIDIA Nemotron Reasoning Challenge — LoRA fine-tuning of Nemotron-3-Nano-30B (Hugging Face, Unsloth, TRL); OpenAI Agent-Security Multi-Step Tool Attacks — LLM agent red-teaming and adversarial prompt injection; Google DeepMind Gemini 3 Pro hackathon (Winner — see Builds).

## Education
- **MD/PhD/MPH — UTMB John Sealy School of Medicine** (expected 2029; PhD defense on track for Dec 2026). USMLE Step 1 passed; advanced to PhD candidacy. Advisor: Peter McCaffrey, MD.
- **BS Biology, minor Chemistry — St. Mary's University** (2021). NCAA tennis student-athlete.

## Certifications
- Monitoring Production Agents — LangChain Academy (Apr 2026)
- Ambient Agents with LangGraph — LangChain Academy (Feb 2026)
- Deep Research with LangGraph — LangChain Academy (Feb 2026)
- Deep Agents — LangChain Academy (Feb 2026)
- Introduction to Agent Observability & Evaluations — LangChain Academy (Jan 2026)
- LangGraph Essentials (Python) — LangChain Academy (Jan 2026)
- Introduction to LangChain (Python) — LangChain Academy (Dec 2025)
- Level 1 Certification — Emerald Cloud Lab (Nov 2024)

## Toolkit
- **Languages:** Python · SQL · R
- **AI & agents:** LangGraph · LangChain · LangSmith · multi-agent orchestration · LLM API integration
- **Retrieval & models:** vector databases · embedding models · on-device LLMs (Ollama) · LoRA fine-tuning
- **Cloud & data:** Azure (HIPAA) · PostgreSQL · Retool · PubMed / NCBI APIs

4+ years of applied AI and natural-language processing; 3+ years building against the PubMed/NCBI APIs.

## Links
- Full CV: https://james.jcweatherhead.com/resume/ (Markdown: https://james.jcweatherhead.com/resume/index.md)
- GitHub: https://github.com/JamesWeatherhead
- LinkedIn: https://www.linkedin.com/in/james-weatherhead-7ab4b9321/
- Email: jamescharlweatherhead@gmail.com

## For agents
- Curated index: https://james.jcweatherhead.com/llms.txt
- Structured profile (schema.org): https://james.jcweatherhead.com/profile.json