I take ambiguous business problems, reduce them to systems that hold up under load, and ship products users stop noticing. The best software disappears. Users remember the outcome, not the interface.
quitefounder.com · LinkedIn · 7Unit · hello@quitefounder.com
Currently building 7Unit, an Operational Engineering & AI Partner for growing product companies.
Also open to Principal, Staff and Lead Engineering opportunities. Remote worldwide, relocation to Dubai possible.
| Problem | What I built | Outcome | |
|---|---|---|---|
| Hally Healthcare infrastructure | Outpatient clinics run on disconnected tools for scheduling, payments and patient follow-up. Every handoff leaks patients and revenue. | An operating platform for outpatient care: appointment orchestration, digital payments, CRM, patient journey and a developer platform with third-party integrations and workflow automation. | Clinics run scheduling, billing and follow-up from one system. Integrations ship through the developer API instead of custom projects. |
| TRACE Privacy infrastructure | India's DPDP Act requires consent lifecycle management and auditability that most companies have no systems for. | A consent and compliance platform: consent capture and withdrawal, immutable audit trails, privacy operations tooling and enterprise integrations. | Consent decisions become auditable records. Compliance moves from spreadsheets and email to a system of record. |
| Fintech platform (client confidential) | Payment and identity flows spread across legacy services with no single source of truth for a transaction. | Transaction orchestration, identity and KYC workflows, and an integration layer for banking partners. | One source of truth per transaction. Reconciliation and partner onboarding stopped being manual work. |
| Raaho Founding engineer | Idea stage, no platform, need to reach production and prove the model before runway ran out. | The technical foundations from zero to one: core platform, data model, deployment pipeline and the first production release. | Platform reached production and became the base the company scaled on. |
Design note: Hally
Design decisions
Drawn from the Hally ADRs and build spec in the gethally repo:
- Sovereign infra over BaaS. Self-managed Postgres, Keycloak, Redis and Celery. Same images run on cloud or hospital on-prem — PHI sovereignty and no vendor lock-in.
- Shared DB, hard tenancy. Postgres RLS plus explicit
tenant_idfiltering on every repository call. Tenant context comes only from a verified JWT. Per-tenant DEKs make DPDP cryptographic shredding possible. - Modular monolith first. One FastAPI app with clear layers; Celery workers are processes, not services. No Kafka or microservice split until a measured bottleneck demands it.
- Encounter-centric workflow. WhatsApp thread, appointment, payment and clinical note share an
encounter_id. Canonical status lives on the encounter; UI steps are views. Remote booking confirms only on webhook-verified payment (or an audited waiver). - Integrations are a product.
/v1/integrations/*with scoped API keys and HMAC-signed outbound webhooks. EMR/HIS stays the clinical system of record; domain commits never wait on EMR acceptance. Mutating routes honorIdempotency-Key. - Journeys as template → assignment → events. Visit stations (and PREP protocols) snapshot a template at assign time; transitions are append-only. In-flight patient promises are never mutated by later template edits.
- Small PRs, trunk-based. Feature flags over long-lived branches. If a PR needs a meeting to review, it's too big.
- Decisions are written down. Every non-obvious architecture choice gets an ADR. Six months later the "why" matters more than the code.
- Boring by default. Postgres unless there's a reason not to. Managed services before self-hosted. New tech has to earn its place.
- Performance and failure are designed, not patched. Latency budgets, observability and failure handling ship with the first version, not after the first incident.
- AI as a pair, not an author. I use AI tooling heavily for exploration, refactors and tests. Design and review stay with me.
- Git history is communication. Atomic commits, conventional messages, signed commits.
git logshould read like a changelog.
TypeScript · Node.js · Java · Spring Boot React · Next.js PostgreSQL · Redis · Docker AWS · GCP AI-native product engineering
Products where AI, healthcare, privacy and developer experience overlap. If you're working on something in that space, or need senior engineering capacity without growing headcount, talk to me.
Quiet by nature. Built quite a few things.
Most people assume QuiteFounder is a typo. It isn't.
The work has always mattered more than being seen.