The da Vinci platform is the world's most-used robotic surgery system.
I built the mobile and web companion apps that surgeons use to track procedures, manage OR schedules, and review outcomes — deployed across thousands of hospitals globally under strict HIPAA and medical-device compliance. Work spanned React Native, Ionic + Angular, and an AWS serverless backend, all designed so that unreliable hospital WiFi never blocks a surgeon in the OR.
The challenge
Operating rooms are the worst environment for a connected app.
WiFi is shielded or overloaded, surgeons can't wait for a spinner, and every byte of PHI is regulated. The legacy web app broke offline, took too long on reconnect, and its auth model couldn't meet HIPAA session rules without frustrating the people who actually use it.
Shielded roomsWiFi you can’t trust
No spinnerssurgeons can’t wait
Regulated dataevery byte of PHI
The solution
Offline-first, not offline-fallback.
Shipped an offline-first architecture where local is the source of truth and sync is a background concern. WatermelonDB provided observable queries and conflict-free merges across devices. OKTA SSO with biometric unlock met HIPAA session requirements without dragging surgeons through passwords at every pause. Everything that touched PHI was encrypted at rest, pinned in transit, and scoped to short-lived sessions.
Local firstBackground syncAudit trail
Architecture
Offline-first, top to bottom.
Offline-first mobile and web apps on top of an AWS serverless backend. Every layer — auth, storage, sync, network — is hardened for HIPAA, and the app is usable with zero connectivity.
Client
React Native
Ionic + Angular
TypeScript
Offline-capable UI
Offline Data
WatermelonDB
Redux Persist
Background sync
Conflict-free merges
Security
OKTA SSO
Biometric unlock
Keychain / Keystore
SSL pinning
Encrypted at rest
Backend
AWS Lambda
API Gateway
DynamoDB
S3 (HIPAA-eligible)
One request, start to finish
Surgeon signs in with OKTA + biometric unlock
Case data is hydrated locally from the encrypted store
Procedures are tracked offline in the OR
Background sync resolves conflicts on reconnect
Outcomes flow into analytics with full audit trail
Decisions
Five calls and why I made them.
Call 01
Offline-first, not offline-fallback
OR networks are unreliable by design — shielded rooms, RF interference, hospital VLANs. Treating offline as the primary mode rather than a fallback kept surgeons unblocked and eliminated entire classes of sync-state bugs.
Call 02
WatermelonDB over plain SQLite
Needed conflict-free sync across multiple devices per surgeon and observable queries for reactive UI. WatermelonDB's sync primitives removed the hand-rolled merge logic that plagued the previous Cordova app.
Call 03
OKTA SSO with biometric unlock on resume
HIPAA dictates session length and forced re-auth. Biometric unlock on every resume kept the friction low enough that surgeons didn't work around security, which is the usual failure mode in clinical software.
Call 04
SSL pinning with a certificate rotation playbook
Medical data over hospital networks can't trust the network. Pinned certs stop MITM — but pinning without a rotation playbook is how apps brick themselves on expiry day. Both had to ship together.
Call 05
AWS serverless for the backend
Small team, global deployment, bursty hospital traffic. Lambda + DynamoDB + S3 kept cost aligned with usage, passed HIPAA attestation out of the box, and let the mobile team ship backend changes without a dedicated ops rotation.
Reflections
What I took away.
Offline-first changed how I design every app since
Before da Vinci, I treated offline as a fallback mode — something to patch in if users complained. Working in ORs taught me that offline should be the primary mode on anything that touches real-world workflows. The moment you flip that switch, entire classes of sync-state bugs disappear, and the app feels faster even when the network is perfect.
Security that gets in the way will be worked around
The usual failure mode in clinical software isn't that security is too weak — it's that it's so friction-heavy surgeons share accounts, write passwords on whiteboards, or leave sessions unlocked. Biometric unlock on resume gave us HIPAA-compliant session rules without the friction, and actual compliance went up because the humans cooperated with the system instead of fighting it.
Certificate pinning without a rotation playbook is a loaded gun
We shipped SSL pinning early to stop MITM on hospital networks. A few months in, I realized we had no plan for the day the cert rotated. Pinning without a rotation runbook is how apps brick themselves on expiry day. Now I treat the rotation playbook as part of the feature, not documentation you write later.
Outcomes
What shipped.
Uptime99.9%Critical surgical tracking reliability
Surgeons100K+Users across thousands of hospitals
HIPAACompliantFull healthcare attestation
OR Efficiency+25%Fewer scheduling conflicts
Stack
Technology
Frontend
React Native
Ionic
Angular
TypeScript
Backend
AWS Lambda
API Gateway
DynamoDB
S3
Security
OKTA SSO
Keychain / Keystore
HIPAA
SSL Pinning
Offline & Sync
WatermelonDB
Redux Persist
Background Sync
Conflict Resolution
Features
What it does
Offline-first architecture with conflict-free sync across devices (WatermelonDB)
OKTA SSO plus biometric unlock (Touch ID / Face ID / fingerprint) with HIPAA-enforced sessions
HIPAA-compliant storage and transport: encrypted local DB, Keychain/Keystore, SSL pinning
Real-time OR procedure tracking with surgeon-annotated timelines
Drag-and-drop OR room management with optimistic updates
AWS serverless backend (Lambda, API Gateway, DynamoDB, S3) with full audit trail
99.9% uptime and observability into sync lag and failure modes
Deployed to 100K+ surgeons across thousands of hospitals globally