yasmin test --career
Yasmin Arfeen
Registered Occupational Therapist, now in AI-first QA.
Agents do the legwork. A person who knows what good looks like decides what passes.
Career test run
4 suites passed
PASSotomate.specOTomateCreatorPrivate beta3 tests
A clinical documentation assistant. Record a session and get a draft note. The clinician reviews it and finishes it.
- Passed: Designed it and wrote the specs
- Passed: Built it by directing AI coding agents
- Passed: Acceptance-tested the builds as its primary user
PASSpaper-and-seal.specPaper&SealLead QA3 tests
A live SaaS for digital event invitations.
- Passed: Lead QA over an AI agent that reviews every branch
- Passed: Catch what automation misses, on real phones and desktops
- Passed: Make the ship call before customers see it
PASSclinical-practice.specClinical practiceRegistered OT3 tests
- Passed: Occupational therapy roles
- Passed: Ran the tech and the therapy in virtual care for northern communities
- Passed: Notes that are accurate, concise and defensible
PASSeducation.specEducation3 tests
- Passed: MScOT, University of Toronto
- Passed: Honours BSc, Health Studies, University of Waterloo
- Passed: Registered with the College of Occupational Therapists of Ontario
- Suites
- 4 passed, 4 total
- Tests
- 12 passed, 12 total
- In progress
- AI-first QA
Why an OT tests software
An assessment and a test plan ask the same question: does this hold up when a real person depends on it?
| In the clinic | In QA |
|---|---|
| Functional assessment | Real-use testing |
| Standardized assessments | Repeatable test cases |
| Measurable goals | Acceptance criteria |
| Risk screening | Risk-based priority |
| Charting | Defect reports |
How I run the suite
A draft is only a draft until a person signs it.
Verify AI output
Agent-written code and tests get checked in the running app.
Criteria first
Every test traces back to something a person can check.
Expert sets the line
The person who knows the work decides pass and fail.
Run it twice
AI is not deterministic. What matters gets run more than once.
Rules before graders
Plain checks where they work, model grading only for nuance.
No real people in tests
Synthetic clients and sample text only.
Get in touch
Hiring for QA, AI evaluation or health tech? I would like to hear from you.
- [Email]
- Phone
- [Phone]
- [LinkedIn URL]