I'm a backend-leaned developer from Nepal currently studying at Chitkara University. I build systems that survive real traffic, not just demo day — and I learn by shipping, not by watching tutorials.

I don't chase perfect systems. I chase systems that know how to break gracefully.

A fault is one component misbehaving — a PostgreSQL connection pool leaking, a Render instance cold-starting for 50 seconds, a cron job firing twice because you forgot to guard it. A failure is the whole platform collapsing because nobody asked "what if?" I design for the first so I never have to explain the second.

Scalability isn't a badge you earn once. It's the question you keep asking — what happens when this grows 10x? I load-tested my job portal at 100 concurrent users when I barely had 5. Not because I needed to, but because the bottleneck should surprise you in data, not in production.

Even AWS doesn't build 100% fault-tolerant systems. They build systems that know which faults matter. That's the standard I hold my work to.

Backend · DevOps · Database Architecture · System Design

Built this solo, from the database schema up. 50+ REST APIs, JWT auth with RBAC across three user roles, and a PostgreSQL schema tuned with composite and GIN indexes that took search latency from 7ms down to 0.9ms.

The backend is strict MVC — controllers, services, models, thin routes. I wrote the migration system as 14 numbered SQL files with dependency ordering (extensions → enums → tables → indexes → triggers) so the schema evolves cleanly. Every table has database-level validation — even if someone bypasses the client and server, the data stays clean.

I didn't just ship features. I tried to break it:

• Load-tested with Apache Bench: 67 req/sec sustained, 100 concurrent users, zero failures
• Multi-stage Docker builds shrank the backend image from 1.99GB to 520MB
• Caught a PostgreSQL pool leak before it hit production — learned that Pool queries must be released or transactions silently fail
• Built a cron job that keeps the free-tier Render instance alive, taking cold starts from 50+ seconds to under 2s — 99.5% uptime over 3 months
• 28+ integration tests with Jest and Supertest covering the critical routes

Also shipped an AI resume scorer using Grok + pdf-parse, a notification system for followed companies, and full Swagger docs.

Tech: Node.js · Express · PostgreSQL · Prisma · Docker · Jest · Supabase · Render

Frontend · State Management · Performance · UX

Drag-and-drop kanban boards with dark mode, keyboard shortcuts, and full responsive layout. The core is a custom undo/redo engine I built with three-stack state history — past, present, future — handling 50+ sequential operations at O(1) without lag.

I used React Context API with localStorage persistence, but added hydration safety checks so refreshing the browser never corrupts state. For the drag-and-drop, I optimized rendering with strategic memoization and ref-based callbacks to keep 60fps even with 50+ task cards mounted. Built client-side CSV/JSON export with Blob generation — zero server dependency.

Tech: Next.js · React · TypeScript · Tailwind CSS · DnD Kit · Chart.js

Full-stack · Authentication · Cloud Integration · Deployment

My first serious backend project. MVC pattern with Express and MongoDB. Session-based auth using Passport.js with salted+bcrypt hashing and a MongoDB-backed session store. I learned the hard way that MemoryStore leaks in production — switched to connect-mongo before deploying.

Built an image upload pipeline with Cloudinary (multer parses, Cloudinary stores, database keeps the reference) and integrated MapTiler for geocoding — type an address, get a live map marker. Deployed on Render with environment isolation.

Tech: Node.js · Express · MongoDB · Passport.js · Cloudinary · MapTiler · EJS

I don't collect LeetCode numbers for the sake of it. I solve problems because every hard question teaches me to see the shape of a solution before I write code.

• 325+ problems solved on LeetCode (peak rating 1570, 365-Day Badge)
• Active on Codeforces — currently solving A & B level problems, building up contest stamina
• Languages: C++ (primary for CP), JavaScript, TypeScript, SQL

• Next.js as a full-stack framework — not just frontend, but wiring APIs directly in the app layer with TypeScript
• Redis — not just "caching," but persistence models, rate limiting internals, and why hash sets perform differently under load
• AWS — hands-on with S3, EC2, IAM, and Lambda
• Clean code as a discipline — writing code so clear that anyone from junior to senior can read it without asking questions

• 🥇 Winner — Code-a-Thon 7.0 (university hackathon, cash prize)
• 🌟 Shortlisted for Round 2 — MyPolaris Fellowship 2026 (selected from 10,000+ students across 300+ cities)
• 🌟 Selected Contributor — GirlScript Summer of Code 2026 (national open-source program)
• 📚 CEED Member — Entrepreneurship Program, Chitkara University

I don't hit every topic above every single day — time is tight. But I touch something meaningful at least biweekly. Consistency over intensity.

Open to backend, full-stack, SDE, and DevOps internships across India.

📫 dipsharmadev@gmail.com · LinkedIn · Portfolio