Rohit Raj
Founding Engineer · AI Systems Architect · India
I'm Rohit Raj — a founding engineer and AI systems architect based in India with 10+ years of experience shipping production software. I specialize in building full-stack applications that combine AI capabilities with robust backend infrastructure. My work spans mobile apps (React Native, Kotlin, Jetpack Compose), AI-powered systems (RAG pipelines, on-device inference, LLM orchestration), and enterprise Java architecture (Spring Boot 3.x, Java 21, Kafka, PostgreSQL).
What I build
I've built multi-tenant SaaS platforms, WhatsApp chatbots for healthcare, AI travel planners, and privacy-first financial advisors. Every project ships with real infrastructure — CI/CD pipelines, observability, load testing, and deployment automation. I don't build demos; I build systems that run in production.
How I got here
Before going independent, I worked on distributed systems and event-driven architectures at scale. I've designed Kafka pipelines processing millions of events, built caching layers that reduced database load by 80%, and architected microservices that handle thousands of concurrent requests.
Why this site exists
I document what I build because engineering work is often invisible. This site is a living record of problems I've solved, architecture decisions I've made, and trade-offs I'm still navigating. Most AI products stall between demo and production. If you're trying to cross that gap, I'd like to hear about it.
Tech stack I ship with
Production-grade tools, picked per project. No framework-of-the-week churn.
Backend & Infra
- Spring Boot 3.x
- Java 21
- Kafka
- PostgreSQL 16
- Redis 7
- Docker
- Flyway
Mobile & Frontend
- React Native + Expo
- Kotlin
- Jetpack Compose
- Next.js 16
- TypeScript
- Tailwind 4
AI & Data
- Spring AI
- OpenAI
- AWS Bedrock
- Gemma 4
- LiteRT
- RAG + pgvector
- LLM orchestration
Every project ships with real infrastructure
Not demos. Observability, load tests, contract tests, and Kafka verification by default.
Stuck between demo and production?
Most AI products stall there. If you're trying to cross that gap, I'd like to hear about it.