Case Study: Tusko - Dental Clinic & Lab Operating System

Project at a Glance

Role: Full-Stack Engineer (end-to-end ownership across API, workers, UI, integrations, deploy)
Duration: Jun 2025 - Present
Live: app.tusko.ca
Stack: React 19, Vite, TypeScript, Tailwind, Express, Prisma, PostgreSQL, Redis, Bull, AWS S3, Docker, Nginx, iTero, Medit
Scale: ~95,000 lines of production code, 39 Prisma models, multi-tenant clinic + lab roles
Team: Small product team - I owned core platform architecture, scanner integrations, case lifecycle, and production deploys

Problem

Dental clinics and labs still run restorative cases across email, shared drives, phone calls, and scanner portals. Prescriptions get lost, files are duplicated, designers do not know what is urgent, and clinics cannot see case status without chasing the lab. Tusko replaces that fragmented workflow with one production system for clinics and labs.

What I Built

Multi-tenant clinic & lab OS

Clinic organizations with members (admin, dentist, receptionist, manager) and preferred-lab relationships
Lab organizations with designers, services, onboarding invites, and referral flows in both directions (clinic invites lab, lab invites clinic)
Case lifecycle as a first-class domain object: create, assign, design, review, deliver, with event history and nudges
Patient records owned by the clinic, optionally linked to cases, with intake token flows for patient data collection

Scanner & file pipeline

iTero OAuth + order sync and Medit webhook ingestion so cases can enter Tusko from real scanner systems
Scanner integration models, clinic mappings, sync jobs, and background file download jobs
Case file storage on AWS S3 with zip handling, viewer UX, and durable worker processing via Redis/Bull

Product surfaces that operators actually use daily

Case board, calendar, deep case detail, designer assignment, lab OS standalone mode for labs importing their own scanner work
Insights/metrics for lab throughput, notifications, estimates, and documentation walkthroughs for clinic setup
Role-aware UI so a clinic admin, designer, and lab admin each see the correct world

Production infrastructure

Docker Compose for staging and production, Nginx reverse proxy with TLS for app.tusko.ca (and mintlabs alias)
GitHub Actions deploy workflows with Slack notifications on success/failure
Seeded roles and environments so clinics/labs can be onboarded repeatedly without tribal knowledge

Architecture Highlights

Clinic / Lab UI (React + Vite)
        |
        v
   Nginx (TLS)  --->  Express API (Prisma + Postgres)
        |                    |
        |                    +-- Auth / RBAC / Clinic-Lab tenancy
        |                    +-- Cases, Patients, Messages, Estimates
        |                    +-- iTero OAuth + Medit webhooks
        |
        +---------->  Redis + Bull workers
                         |
                         +-- Integration sync jobs
                         +-- File download / S3 pipelines
                         +-- Notification fan-out

Hard Problems I Solved

Two worlds, one product: Clinic OS and Lab OS share models but must not leak tenancy. Designed ownership rules so lab-imported scanner cases can exist without a clinic account, while clinic-owned cases still route cleanly to preferred or external labs.
Scanner reality is messy: iTero and Medit payloads, OAuth, and mapping are not clean REST. Built durable sync jobs, mapping tables, and worker retries instead of fragile request/response assumptions.
Case state without chaos: Cases accumulate files, messages, estimates, feedback, and designer handoffs. Event + nudge models keep the timeline reconstructable and actionable.
Onboarding that sells itself: Referral links, lab requests for labs not yet on Tusko, and guided clinic/lab onboarding so growth is productized, not spreadsheet ops.

Results & Impact

Production platform live at app.tusko.ca used by real clinics and labs
Full restorative case loop in one system: patient -> prescription/files -> lab/designer -> delivery
Scanner-connected intake reduces manual file chasing
Deployable staging/prod path with Docker + Nginx + CI

What This Shows About Me

I can own a complex B2B domain (healthcare-adjacent ops) from schema to UI to workers to production
I design multi-tenant RBAC and lifecycle systems that survive real operational edge cases
I integrate third-party industry systems (iTero/Medit) with durable jobs, not demos
I ship production infrastructure, not just feature branches