A primary-care telehealth startup needed a platform that satisfied HIPAA (US) and ABDM (India) on day one — not an after-thought. We delivered a patient app, doctor app, and EHR core that now holds 80,000 patient records and processes 14,000 consults a month.
Indian telehealth has a trust ceiling. Patients won't book a video consult unless they trust the credentials, the prescription is honored at a pharmacy, and the data does not end up on a billboard. The client wanted to build a clinic-grade telehealth product — not a glorified video-call app.
We split the platform into three products plus a compliance-aware data layer. The non-obvious choice was making compliance a service (the audit + consent layer) instead of a checklist; everything that touches patient data routes through it.
Symptom checker, doctor search by specialty + language + price, video consult, prescription view + share, lab booking, pharmacy fulfillment, appointment history, family profiles. ABHA linking on first launch (with skip-for-now).
Schedule view, queue, video call, ICD-10 prescription builder, voice-to-text notes, e-signature, payout dashboard. The prescription builder is the core daily-use surface — we iterated on it five times.
Patient record view (FHIR-compliant), consent log, audit trail explorer, MCI/NMC verification queue, payout reconciliation, ABDM consent-manager dashboard, support desk.
Every read/write to patient data goes through a service that enforces consent, encrypts at rest with KMS, redacts PHI in logs, and writes to an immutable audit table. Penetration-tested by a third-party CERT-In auditor.
TypeScript-first, FHIR-compliant patient model, ABDM HIE-CM + HIP integration, encrypted-at-rest with KMS, encrypted-in-transit, BAA-signed AWS region (Mumbai), DLP-monitored logs.
We picked the stack for fit, performance under realistic load, and operational simplicity. Here is the breakdown:
NestJS over Express: the dependency-injection + module pattern made the consent-and-audit middleware actually clean. Healthcare code where every layer needs an audit log is exactly where Express becomes a maintenance liability.
FHIR R4 from day one: not because the client asked, but because retrofitting FHIR is misery. Every patient record is FHIR-compliant; integrating with a hospital partner six months in took three days, not three months.
AWS HIPAA-eligible region: we ran in Mumbai with a BAA signed. Every service we used (RDS, S3, KMS, CloudFront) is on the BAA-eligible list. This is the part most Indian dev shops get wrong — it is not enough that AWS supports HIPAA; you have to use only the eligible services.
ABDM consent-manager flow.
ABDM's HIE-CM (Health Information Exchange & Consent Manager) flow is the hardest engineering problem in the whole stack — and the most rewarding once it works, because that is what makes prescriptions credible at any partner pharmacy and lab.
The challenge: a consent request is asynchronous, multi-party, and revocable. The patient initiates, the consent manager (a trusted third party) verifies, the requester (us) waits, the provider (us, in a different role) responds, and the entire trail is signed.
We modelled it as a state machine with 11 states and 23 transitions, persisted to Postgres with optimistic locking and replayed via a job queue. Every state change is appended to an immutable audit table with a digital signature.
Time to first consent in production: 6 weeks. Time to ABDM-sandbox-pass: 4 months. Worth it: this is the moat. Competitors who skip ABDM cannot get prescription revenue. Period.
We piloted with 12 hand-picked doctors in two specialties (general medicine and dermatology) for eight weeks. The product looked nothing like v0.4 by the time we expanded supply.
Specialty rollout post-pilot was driven by a config-first specialty engine: adding a specialty is a JSON file plus a custom intake form. Six new specialties shipped in seven weeks, two of them (mental health, paediatrics) needed bespoke flows we had not anticipated.
“ITD got the boring stuff right — the consent flow, the audit trail, the BAA region. That is what investor diligence cares about. The fact that the patient app is also one of the best-rated in the category was a bonus.”
Founder & Chief Medical Officer
Telemedicine Platform · India
Invest in clinical-content review earlier. We launched the symptom-checker copy with a paid panel of generalists; the dermatology vertical's content needed dermatology-specific review, which we discovered in user testing post-launch. Specialist content review at v1 = no embarrassing moments at v1.5.
Compliance is not a checklist; it is the architecture. We have shipped HIPAA + ABDM + DPDP-compliant healthcare platforms — patient apps, EHRs, hospital integrations. Talk to an engineer who has done it.
Get a Free Consultation
