Backend and full-stack engineer from Dhapasi, Kathmandu — BSc CSIT at Amrit Campus, building real production systems while studying.
I'm not chasing tutorials. Right now I'm building:
- 🍱 The Mummy Sweets & Corner — a full restaurant platform with production-quality auth, background jobs, and RBAC
- 🆘 Sahayog — AI disaster-response + geospatial mutual aid system
I'm using these to understand why systems are designed the way they are — not just how to make them work. Service/Repository architecture, async patterns, token refresh races, database query planning — this is the level I operate at.
💀 Question I ask before every line of code: What happens when this fails at 3AM under concurrent load?
📍 Dhapasi, Kathmandu · 📬 sharmashekhar20050@gmail.com · 📞 +977-976-557-1937
ram = {
"currently_building": [
"🍱 The Mummy Sweets and Corner — production restaurant platform",
"🆘 Sahayog — AI disaster response + geospatial mutual-aid system",
],
"architecture": "Service / Repository / Schema pattern",
"backend": {
"primary" : ["FastAPI", "async/await", "Pydantic v2", "Dependency Injection"],
"orm" : ["Async SQLAlchemy", "Alembic migrations"],
"deepening" : ["Django", "Django REST Framework"],
"infra" : ["PostgreSQL", "Redis", "Celery", "Docker / Compose"],
"auth" : ["JWT", "HttpOnly cookies", "access + refresh tokens", "RBAC"],
"storage" : ["Cloudinary", "background image processing via Celery"],
"observability": ["structured logging", "error handling middleware"],
},
"frontend": {
"core" : ["Next.js 14 App Router", "React", "TypeScript", "TailwindCSS"],
"data" : ["TanStack React Query", "Zustand", "Axios interceptors"],
"forms" : ["React Hook Form", "Zod"],
"ui" : ["shadcn/ui", "Framer Motion"],
},
"previously_shipped": [
"Multi-vendor e-commerce (Stripe webhooks, Inngest, OpenAI)",
"AI Resume Builder (PERN + OpenAI)",
"Next.js 14 Blog (NextAuth, RBAC, ISR caching)",
"Expense Tracker (MERN, Chart.js, JWT)",
],
"engineering_thinking": [
"Why access tokens expire → refresh tokens exist → where each lives",
"10 requests hit 401 simultaneously → refresh race → refreshPromise",
"PostgreSQL EXPLAIN ANALYZE → index strategy → query planning",
"Redis as cache vs broker → TTL → distributed locking → rate limiting",
"Celery retry logic → idempotency → task failure handling",
"Server Components vs React Query → hydration/dehydration boundary",
"Service layer vs Repository layer → why the separation matters",
"Pydantic v2 validation → dependency injection → lifespan events",
],
"leveling_up": [
"PostgreSQL query optimization + EXPLAIN ANALYZE",
"Observability — structured logging, tracing, Prometheus",
"Testing — pytest, async test patterns, integration tests",
"CI/CD — GitHub Actions, Docker registry, zero-downtime deploys",
"PostGIS + geospatial queries (Sahayog)",
"Kafka + event-driven architecture",
"Distributed systems — consistency, fault tolerance",
"Kubernetes — because Compose only takes you so far",
],
"philosophy": "Ship real things. Understand what breaks. Build for the next order of magnitude.",
"status" : "🟢 Open to backend, full-stack, or AI engineering roles",
}
📁 Service / Repository Architecture (what I actually follow): Request → Router → Dependency Injection ↓ Service Layer (business logic) ↓ Repository Layer (DB queries only) ↓ Async SQLAlchemy + PostgreSQL ↓ Pydantic v2 schemas (in/out/db)
🔐 Auth flow I'm building:
Login → FastAPI ├── Access Token → response body (short-lived, 15min) └── Refresh Token → HttpOnly cookie (long-lived, 7d) ↓ Frontend: Access token in Zustand memory (never localStorage) ↓ Axios attaches Bearer token to every request ↓ 401 → response interceptor fires ↓ refreshPromise ← ONE shared promise (solves concurrent races) ↓ New access token → retry ALL queued requests simultaneously
⚡ Next.js + React Query data boundary:
/menu, /product/[id] (SEO-critical public routes) ↓ Server Component fetches → initialData passed down ↓ React Query hydrates → ZERO double fetch ↓ Interactions: cart, filters, pagination, mutations → React Query owns it → Optimistic updates
🗄️ Database problem-solving approach:
Business problem → Raw SQL → EXPLAIN ANALYZE ↓ Index strategy → SQLAlchemy ORM → Prod at scale
|
Full restaurant platform — built solo, end-to-end. Service/Repository architecture. Production auth with HttpOnly cookies, refresh-token concurrent-request race handling via |
AI-powered disaster-response + community mutual-aid platform. Incident reporting, volunteer matching, blood donors, geospatial shelter lookups, AI-assisted prioritization with coordinator approval. My gateway into event-driven architecture, PostGIS, and Kafka. |
|
Full marketplace — Stripe webhooks, Inngest background jobs, AI product descriptions, admin analytics. Processes real orders.
|
AI resume generator — real accounts, templates, real-time preview, PDF export, ATS hints. Built because I needed one. Stayed because others did too.
|
-%2300ff88?style=flat-square&labelColor=0d1117){kind=icon}
Open to backend, full-stack, or AI engineering roles.
I can build and ship. What I'm actively building right now is engineering judgment — the ability to look at a system and think:
What is the actual problem? → What are my options? What are the trade-offs? → What happens under failure? What happens at 10x scale? → What architecture makes sense?
I'm not just learning FastAPI. I'm learning why Service/Repository separation exists, why Pydantic v2 dependency injection beats global state, why a refreshPromise solves concurrent token refresh races, why EXPLAIN ANALYZE matters before you think indexes.
I want to work somewhere with a problem worth owning. Where the architecture has real reasons. Where code review is a conversation, not a rubber stamp.
I'm not looking for a job. I'm looking for a team with a problem worth solving.
Let's build something real.
