Medical Coding - Coding AI Assist (SME / Product Enablement)

About this role

About Contiinex
Contiinex is a Bengaluru-headquartered technology company developing AI-powered solutions for the US healthcare Revenue Cycle Management (RCM) industry. Our solutions support and automate workflows across patient access, eligibility, prior authorization, medical coding, claims, denials, AR follow-up and collections. We work to convert real RCM operational expertise into practical, auditable software for healthcare providers and RCM service organizations.
The Opportunity: Build Coding AI Assist
We are hiring an experienced US Medical Coding Trainer to serve as the primary coding-domain expert for the design, development, training and validation of Contiinex Coding AI Assist.
This is not a routine classroom-training or production-coding position. You will convert real-world coding judgment into detailed specialty scenarios, coding rules, reference decisions, workflows and test cases that software engineers and AI teams can implement and evaluate.
The long-term product objective is to progress from coding assistance and human-reviewed recommendations toward safe, measurable autonomous coding for explicitly approved use cases. Coding decisions must remain grounded in complete documentation, applicable code-set versions, payer rules and appropriate human oversight. No programming or machine-learning experience is required.

Requirements
Key Responsibilities
Coding Knowledge to Engineering Requirements

• Own the coding-domain requirements for Coding AI Assist: describe input documents, clinical facts to extract, outputs, decision steps, exceptions, evidence, confidence or uncertainty, and when the system must defer to a human coder.

• Prepare step-by-step specialty coding workflows covering chart review, diagnosis selection and sequencing, procedure selection, E/M assignment, modifier application, medical necessity, claim edits and required documentation.

• Build a structured scenario library that includes routine, complex, ambiguous, incomplete, conflicting and rare cases; multiple diagnoses and procedures; post-operative care; global periods; add-on codes; exclusions; and coding changes over time.

• Write understandable rule specifications and decision trees for engineers, including trigger conditions, required documentation, exact expected coding decision, rationale, references, version and date applicability, exceptions and test acceptance criteria.

• Create de-identified or properly authorized example charts and synthetic scenarios with gold-standard coded answers; coordinate independent review and adjudicate disagreements with documented reasoning.

• Work directly with product managers, engineers, data and AI specialists and QA teams in requirement workshops, sprint reviews and defect triage. Explain clinical and coding concepts without assuming technical coding knowledge.

• Define which decisions are safe to automate, which require provider clarification or a compliant non-leading query, and which require specialist review rather than a guessed code.

• Maintain change-control documents as ICD-10-CM/PCS, CPT, HCPCS, official guidelines, NCCI, MUE and relevant payer and policy requirements change.
AI Training Data, Documents and Knowledge Artifacts

• Develop a medical coding taxonomy and specialty-by-specialty coverage matrix: encounters, services, procedures, diagnoses, terminology, document types, modifiers, billing contexts and common exception patterns.

• Author a coding rules repository, clinical-documentation checklist, scenario catalog, decision-tree library, annotated sample charts, reference coding dataset, query templates, error taxonomy and release and version history.

• For each case, capture source evidence, correct codes and sequencing, alternatives rejected and why, guideline or policy citation, effective date, expected AI output and escalation conditions.

• Build balanced training, validation and holdout test sets with varied note quality, clinical complexity, specialty, code frequencies and edge cases; prevent case leakage between evaluation sets.

• Define acceptance tests and measurable quality metrics: code-level accuracy, principal or first-listed diagnosis accuracy, E/M and modifier accuracy, unsupported-code rate, missed-code rate, escalation quality and specialty-specific performance.
AI Tool Testing, Evaluation and Continuous Improvement

• Run hands-on user acceptance testing (UAT) on new and updated Coding AI Assist builds; compare suggested codes and rationales against adjudicated reference answers.

• Identify overcoding, undercoding, unsupported specificity, missing codes, incorrect sequencing, modifier misuse, NCCI conflicts, E/M level errors, medical necessity issues and fabricated facts.

• Log reproducible defects with chart context, expected vs actual output, severity, guideline or reference, specialty, root-cause hypothesis and recommended correction.

• Retest fixes and regression suites across specialties and code-set versions; provide evidence-based go/no-go recommendations and monitor performance after release.

• Train internal teams and pilot clients on appropriate AI review, uncertainty, exception handling and safe use of the tool.
Specialties and Settings to Be Covered
The product roadmap aims for broad coverage across the following areas. One trainer is expected to own the overall scenario framework and prioritize coverage; expert validation or additional specialty SMEs will be used where depth or certification is needed.

• Primary care and ambulatory: family medicine, internal medicine, general practice, geriatrics, pediatrics, adolescent medicine, preventive medicine, occupational medicine, urgent care and telehealth.

• Medical subspecialties: cardiology, interventional cardiology, electrophysiology, pulmonology, sleep medicine, gastroenterology, hepatology, nephrology, endocrinology, rheumatology, allergy and immunology, infectious disease, hematology and oncology.

• Neurology and behavioral health: neurology,…