In this role you'll define the clinical standard our AI is measured against — building gold-standard datasets, validating model output before it reaches a transplant center, and owning center go-live end to end. You're the clinical conscience of our automation: the person who turns registry reality into labeled truth, catches failure modes before they ship, and earns enough trust that centers stake accreditation on AI-abstracted CIBMTR output.
You'll also own clinical QA for our upcoming hybrid quantum ML (HQML) paths — the same registry-grade bar, structured classical-vs-HQML evaluation, and explicit go/no-go criteria before any change enters production workflows.
This is a hands-on, high-trust founding role at the intersection of clinical AI validation, quality assurance, registry operations, and customer success — one of our first hires and the clinical backbone of the team.
The Team
We're a small, deeply technical founding team turning the high-stakes manual workload of stem cell transplant clinical data abstraction into seamless AI automation — and building the quantum-enhanced analytics and prediction layer on top of structured clinical data. Our co-founders bring years running stem cell transplant clinical data operations and deep expertise in ML and quantum computing. We work with the largest cancer centers that define this field. The decisions our data informs sit behind cutting-edge medical procedures and determine whether patients live.
In This Role You Will
- Own clinical gold standards: Build and maintain human-abstracted ground truth across CIBMTR forms — the evaluation set every model release is scored against.
- Validate before you ship: Review extraction output against gold labels; adjudicate disagreements; document failure modes and turn every miss into a labeled example that improves the system.
- Benchmark hybrid automation: Run structured classical-vs-HQML QA on agreed metrics — per-field accuracy, rare-field recall, auto-fill rate, and review burden — before promoting any path to centers.
- Govern center go-live: Lead transplant-center onboarding end to end — kickoff, data mapping, workflow fit, training, and hands-on support that turns skeptical registry teams into advocates.
- Translate abstractor reality into product: Convert how registry teams actually work into requirements our engineers and model teams can build against.
- Partner on clinical AI governance: Work with founders on intended-use boundaries, escalation paths, rollback triggers, and the quality evidence centers need to adopt automation.
- Defend registry-grade quality: Own the edge cases — rare diseases, messy regimens, ambiguous fields — that separate audit-ready CIBMTR data from "close enough."
What We Hope You'll Bring
- Hands-on CIBMTR abstraction experience — you've personally completed Forms 2400, 2402, and/or 2450 and know where they get hard.
- Deep familiarity with transplant/cellular-therapy clinical data and registry reporting workflows.
- A clinical AI / clinical data QA mindset — obsessive about correctness, edge cases, and evidence, not just throughput.
- Credibility with center data teams and clinicians; clarity with engineering and product.
- Comfort in an early-stage environment: high ownership, little process, things change weekly.
Bonus Points If You Have
- Previous NMDP, CIBMTR, or BMT CTN experience.
- Experience leading or training a registry data team.
- Familiarity with FACT accreditation requirements.
- Built or maintained clinical evaluation datasets, annotation workflows, or gold-set regression for NLP/LLM systems.
- Exposure to SQL, structured data QA, or annotation platforms — or a real appetite to learn them.
- Experience with human-in-the-loop review, champion/challenger testing, or pre-deployment clinical AI validation (a plus in this role, not a hard filter).
To express interest in this role, email careers@summithealthdata.com with your background and the role title in the subject line.