Work

Proof of work

Documents, research, and public work. Every file below is a real deliverable, opened as it was written. Where something is a worked example rather than a client engagement, it says so.

Samples

PDF

Medical content review, start to finish

A clinical accuracy review of a live, published supplement-brand article. Every claim checked against the evidence, three high-risk interaction omissions found, and corrected language written for each one.

Content reviewClaims checkingPatient safety

PDF

Testing a general model on clinical safety

A battery of clinical-safety prompts covering drug dosing, red-flag symptoms, crisis response, contraindications, and vague-symptom triage, run against a general-purpose model and scored against a fixed rubric. Results reported exactly as measured.

AI evaluationMethodologySafety

PDF

Before and after: an energy supplement article

A worked example for a wellness brand. The original article, the clinical and regulatory findings, and the corrected version side by side. Brand redacted, findings reproduced as delivered.

Fact-checkingRegulatory riskEditing

PDF

Fixing AI-generated patient-facing health content

The original draft, what was wrong with it — generic phrasing, stock transitions, missing specificity — and the edited version, with every correction listed.

AI contentHealth communicationEditing

Interactive

Deploy Check

Eight health-AI products pitched for low-resource clinical settings. You make the call a buyer makes on each one and see the failure class behind every verdict. Composite scenarios built from documented failure modes.

InteractiveDeployment risk

PDF

Falsely flagged by an AI detector? A free guide

For students wrongly accused of AI-written work: how detectors reach a verdict, the published error rates that undermine them, and how to open an informed conversation with a department. Free, no email required.

AI detectionFree resource

Recognition