I design and build backend systems that are resilient under load, safe with data, and predictable in failure.
My work spans multi-tenant SaaS platforms, asynchronous distributed architectures, real-time event systems, and cloud infrastructure on AWS and GCP. I care deeply about correctness, data integrity, and observability not just shipping features.
I think in transactions, queues, and service boundaries. Every system I build is designed to fail safely, scale predictably, emit useful signals when something goes wrong, and be maintainable six months from now by someone who isn't me.
| Domain | Focus Areas |
|---|---|
| Backend Services | NestJS 10 Β· Fastify Β· Django |
| SaaS Architecture | Multi-tenant isolation Β· RBAC Β· Row-level tenancy |
| Async Processing | Redis Β· BullMQ Β· Pub/Sub event pipelines |
| Data Layer | PostgreSQL schema design Β· Transaction safety Β· ACID guarantees |
| Real-Time Systems | WebSocket streaming Β· Event-driven communication |
| Cloud Infrastructure | AWS Β· GCP Β· Containerized production deployments |
| Observability | Grafana dashboards Β· Metrics Β· Structured logging Β· Alerting |
|
Multi-Tenant Medical Shop SaaS A production-grade B2B SaaS platform built for small medical shops across India. Designed with correctness and data integrity as first-class concerns β not bolted on after the fact. Architecture Highlights
|
Async Distributed Processing System A high-performance order execution backend built with Fastify and a Redis-based worker architecture. Designed for throughput, responsiveness, and full real-time observability. Architecture Highlights
|
Design for scale before optimizing prematurely.
Transactions are the first line of defense for business logic.
Validation belongs at system boundaries β not scattered throughout.
Data models determine long-term system quality.
Services must fail safely, loudly, and observably.
Async processing reduces coupling and improves resilience.
You can't fix what you can't measure β instrument everything.
Building backend systems that survive real-world usage.

