Metric Health

The problem they were stuck in

Stood up the technical organization, architecture, and compliance posture for a digital health platform that replaces paper-based clinical evaluation with remote, AI-assisted workflows — serving patients, clinicians, clinic admins, and super admins on a single multi-tenant system. The bet: clinical assessment in community health, rehab, and private practice still runs on paper. That's slow, lossy, and impossible to scale a clinician's panel against. The thesis was a remote, automatically-scored, HIPAA-compliant assessment platform that surfaces the patients who actually need attention — compressing clinician workload while improving clinical signal — if you can earn the trust of regulated buyers from day one. That last clause is the hard part. It set the entire technical agenda.

The shift

• Technical strategy & architecture for a four-portal, role-segmented platform on AWS Amplify Gen 2 (AppSync GraphQL + API Gateway REST, DynamoDB, Lambda, S3, Cognito).
• Compliance posture from day one — HIPAA-compliant infrastructure with encryption at rest and in transit, SOC 2 in active audit, verify-then-trust identity model.
• Security architecture: passwordless auth via WebAuthn passkeys, email OTP, and traditional fallback with device remembering. Invite-only onboarding across all clinical roles.
• AI integration strategy: AWS Bedrock + Google Gemini for assessment summaries and recommendation logic, scoped narrowly to clinician decision support — not autonomous clinical action.
• Engineering org & delivery: built the team, codebase conventions, and testing discipline (Jest + Playwright + SonarQube) that lets a small team ship into a regulated environment without slowing down.
• Go-to-market enablement: shipped a public, no-login ROI calculator that converts top-of-funnel for clinics before sales ever touches them.

The outcome

The system that resulted: • Assessment engine — standardized questionnaires and motor/cognitive tasks with norm-based scoring and flexible scheduling (one-time, recurring, multi-phase). • Longitudinal symptom diary with trend visualization and symptom-driven assessment recommendations. • Clinician action queue — automatic out-of-range flagging with color-coded alerts; referrals, follow-ups, and resource sharing inline. • Export — CSV, PDF, and patient-grouped PDF for clinical and compliance use. • Bulk assignment + flexible scheduling — a single clinician now manages a materially larger panel.

Wins along the way

• Replaced paper assessment workflows with same-device remote evaluation across the entire patient base.
• AI-generated summaries cut clinician review time per patient.
• Public ROI calculator generates qualified self-serve inbound from community health centers, rehab centers, and private practices.
• Built the architecture and compliance scaffolding the company needs to sell into enterprise health systems — not retrofitted later.