
Kelly Silk
Registered Midwife | Domain Expert Clinical AI Trainer
Hire this person
Send a job offer directly to this candidate
About
Registered healthcare professional with 15 years of clinical experience across midwifery and sexual health nursing, and 8 months of specialist AI training work at domain-expert level. I combine deep clinical knowledge with hands-on experience designing the evaluations, rubrics, and complex scenarios that help make frontier AI models safer and more useful. I operate exclusively at the domain-expert tier: assessing clinical reasoning, identifying safety gaps in model outputs, and creating sophisticated, multi-layered prompts that test model functioning and understanding.
My background in clinical guideline development, practice audit, and professional midwifery advocacy has supported me to adapt to the structured, evidence-based evaluation that high-quality AI training requires.
Experience
Domain Expert Clinical AI Trainer at DataAnnotation (2025-10 – Present)
Working exclusively on specialist clinical AI training tasks. Key responsibilities and achievements include:
- Designed complex, specialty-specific prompts to stress-test AI model clinical reasoning in midwifery, obstetrics, and sexual health to deliberately identify gaps in model knowledge and safety handling
- Evaluated AI-generated responses to advanced medical scenarios for clinical accuracy, safety-appropriateness, and adherence to evidence-based guidelines, providing structured written justifications for all assessments.
- Developed rubrics and evaluation frameworks for assessing clinical content quality.
- Conducted complex clinical world-building; constructing intricate multi-variable scenarios from scratch to test a model's ability to track, apply, and remain consistent across lengthy, clinically detailed prompts.
- API and system instruction development.
- Produced detailed written feedback and comparative assessments across multiple model outputs, articulating clinical rationale clearly.
- Conducted systematic safety red-teaming across multiple projects specifically designed to induce model failures, using adversarial prompting techniques, clinical edge cases, and carefully constructed scenarios to expose unsafe outputs, hallucinated clinical information, and failures in appropriate risk escalation within healthcare contexts.
Sexual Health Nurse at British Pregnancy Advisory Service (BPAS) (2022-05 – 2026-02)
- Direct clinical work with clients in abortion care and sexual health, including remote consultation and evidence-based decision support
- Maintained rigorous clinical standards and documentation in a remote, autonomous working environment
Professional Midwifery Advocate (PMA) at British Pregnancy Advisory Service (BPAS) (2019-09 – Present)
Senior leadership role supporting staff wellbeing, driving performance improvement, and leading quality improvement and clinical audit initiatives across the organisation
- Evaluated clinical practice against established standards and produced structured recommendations — skills directly transferable to AI output evaluation
Hybrid Midwife (Birth Centre & Community) at Gloucestershire Hospitals NHS Foundation Trust (2018-12 – 2022-05)
- Managed an independent caseload of pregnant clients; attended out-of-hospital births; conducted triage assessments with autonomous clinical decision-making
Rotational Midwife at Gloucestershire Hospitals NHS Foundation Trust (2017-08 – 2018-12)
Rotated across Delivery Suite and Postnatal Ward, developing breadth of acute and postnatal clinical experience
Caseloading Midwife at One to One Midwives (2014-01 – 2017-07)
Independent community midwife; attended homebirths and hospital births across a personal caseload within a midwifery group practice
Registered Midwife (Band 5 → Band 6) at The Shrewsbury and Telford Hospital NHS Trust (2010-12 – 2014-01)
Progressed from Band 5 to Band 6 within the first year; worked across Delivery Suite, Birth Centre, and Community
Education
BSc (Hons) in Midwifery – University of Worcester (2007 – 2010)