Clinical AI Data Specialist

The Data Science / Clinical AI function is seeking a Clinical AI Data Specialist to ensure the clinical accuracy of the training data, model output labels, and clinical logic — prompts and coding rules — that shape how our AI-powered risk adjustment products behave. This is a clinical coding domain-expert role first: it requires active coding credentials and the ability to independently read, interpret, and annotate clinical medical record documentation, and that expertise translates directly into measurable model performance. Errors introduced at this layer propagate into training and produce systematic clinical inaccuracies at production scale, so the quality of your judgment is the product. The technical work — annotation at scale, prompt and rule iteration, and label-quality analysis — is carried out using AI-assisted development tools; we will train the right clinical coding expert on the tooling, and a software engineering background is not required.

• →      Annotate medical records for AI training data
• →      Validate annotated data to ensure quality
• →      Refine the clinical logic behind AI outputs 
• →      Provide clinical coding & HIM subject-matter expertise to data science

• Read and interpret clinical documentation — physician notes, assessment and plan sections, problem lists, medication records — to identify codeable diagnoses, conditions, and other clinical entities (document boundaries, type, author, section), applying ICD-10-CM and risk adjustment coding standards and mapping to clinical ontologies (ICD-10-CM/PCS, CPT, RxNorm) when required by project scope
• Distinguish conditions that meet documentation standards for coding from those that do not, exercising clinical judgment independently, and flag ambiguous or edge-case documentation with written rationale}
• Review AI model output labels against clinical documentation to identify false positives, false negatives, and specificity errors; clean and correct label datasets and categorize error patterns for the data science team
• Apply coding knowledge to evaluate whether model-generated code assignments are clinically and regulatorily supportable, and escalate systematic quality issues that may indicate model behavior problems
• Translate ICD-10-CM and coding guideline requirements into explicit, testable instructions — LLM prompt language and computable coding rules — using AI-assisted tools testing revisions against curated ground-truth datasets and iterating on observed failures
• Document the clinical rationale and precision/recall impact of each prompt or rule change for senior review

• Domain expertise with a minimum 5 years of coding and/or CDI experience with demonstrated proficiency in ICD-10-CM code assignment from clinical documentation
• Active credential in at least one of: CCS, CPC, CRC, CDIP, CCDS, or equivalent AHIMA/AAPC certification
• Ability to apply clinical coding standards consistently and independently to produce high-quality, reproducible labels across large document sets, catching subtle distinctions that affect code assignment
• Ability to articulate the clinical rationale behind a labeling decision in writing for QA and audit, and to express coding requirements as explicit, unambiguous instructions — the discipline behind a well-constructed coding query
• Works independently within established guidelines without case-by-case direction on routine annotation, and escalates systematic issues — repeated error patterns, guideline gaps, documentation quality trends — rather than resolving them in isolation

• Coding Audit and/or Compliance Experience
• Clinical annotation or AI/ML data labeling experience in a health-tech or healthcare AI environment
• Familiarity with HCC reimbursement models
• Exposure to NLP or ML model outputs in a clinical context — how model-generated codes differ from human-assigned codes