Medical Director, AI Safety

The Role

We are seeking a Medical Director, AI Safety that will own the safety of Avo's AI-generated clinical content. That means deciding what "safe enough to ship" means for a discharge summary, a pre-visit summary, or a decision-support module, building the tests that enforce it, and signing off before it reaches a clinician. You will design the grading rubrics, run the evaluations, classify failures by clinical severity, and turn the findings into changes the Studio and engineering teams make.

Roughly 85% of your time goes to safety. The remaining 15% splits two ways: building clinical modules yourself in Avo's Studio, so your safety judgments are grounded in how the content is actually produced, and joining sales and customer conversations as the physician who can explain how we test and why a CMIO should trust the output.

You report to the CMIO, with a dotted line to the Chief Commercial Officer.

What You'll Do
AI safety (core of the role)

Define the safety standard for each product line: what counts as a hallucination, an omission, or a clinically material error, and how severe each is
Design and run evaluation protocols for clinical AI modules, including hallucination detection, omission analysis, and accuracy checks against source records and evidence-based guidelines
Build and maintain a clinical harm taxonomy and grading rubric (NCC MERP-style severity scoring) and apply it across benchmark sets
Review AI-generated clinical content directly, and grade the automated graders so we know when to trust them
Own the release gate: write the go/no-go safety assessment for new modules and major prompt or model changes
Investigate reported clinical errors from customers, trace root cause, and drive the fix through Studio and engineering
Write the safety reports, validation summaries, and white papers that buyers and their clinical governance committees ask for
Track the regulatory and standards landscape for clinical AI (FDA, ONC, CHAI, Joint Commission) and translate it into internal requirements

Clinical builder (~5-10%)

Build or revise clinical modules in Avo's Studio, applying evidence-based guidelines and your own practice experience
Feed what you learn from building back into the safety protocols, and vice versa

Sales-forward clinician (~5-10%)

Join prospect calls and pilot evaluations as the physician who can speak to how Avo's content is validated
Present the safety program to CMIOs, CIOs, and clinical quality leaders, and answer their hardest questions about failure modes
Challenge the status quo in the room: name where legacy documentation and CDS workflows fail patients, and show how a tested AI alternative does better

What You Bring
Active clinical license (MD/DO, NP, or PA) with hands-on patient care experience
3-7 years of clinical practice; you can read a discharge summary and spot the missed anticoagulation plan in under a minute
Experience with or strong interest in clinical AI, CDS, or health informatics
A testing mindset: you like designing the rubric, arguing about edge cases, and finding the error nobody else caught
Ability to write a clear, defensible safety report for a clinical governance committee
Comfort presenting to C-suite clinical leaders and answering pointed questions about AI failure modes
Willingness to define the role's scope as the product and company grow