I build software in three tiers, and I think the boundaries between them are the most important lines in a codebase. This README is organised the same way.
╔═══════════════════════════════════════════════════════════════╗
║ ║
┌────╨────────────────────────────────────────────────────────┐ ║
│ TIER 1 · PRESENTATION │ ║
│ React · Next.js · Tailwind · Vite │ ║
│ routing, state, rendering, accessibility │ ║
└────╥────────────────────────────────────────────────────────┘ ║
║ typed contracts, and no business logic crosses this line ║
┌────╨────────────────────────────────────────────────────────┐ ║
│ TIER 2 · APPLICATION │ ║
│ Node.js · Express · .NET · Python │ ║
│ validation, auth, services, orchestration │ ║
└────╥────────────────────────────────────────────────────────┘ ║
║ parameterised queries, and no raw SQL above this line ║
┌────╨────────────────────────────────────────────────────────┐ ║
│ TIER 3 · DATA │ ║
│ MySQL · SQL Server · MongoDB · SQLite · Prisma │ ║
│ schema, indexes, migrations, transactions │ ║
└────╥────────────────────────────────────────────────────────┘ ║
║ ║
╚═══════════════════════ runs on YOUR machine ══════════════════╝
I am a Computer Science student at UET Main Campus, Lahore. Most of what I ship is local first, meaning no accounts, no telemetry and no cloud bill. The data never leaves the box it was created on.
What I care about here: rendering that stays predictable, state that stays where you put it, and interfaces that survive a slow connection.
|
A self hosted web toolkit of 43 utilities for images, PDFs and everyday conversions. All of it is processed in the browser. Nothing is uploaded to a server, because there is no server.
|
A zero setup portfolio generator. Enter any GitHub username and get a shareable page instantly, with no auth and no database behind it.
|
Where I actually live. Validation, auth boundaries, background work, and error paths that tell the truth about what went wrong.
|
A dev server control center for Windows. Point it at a folder and it finds every project, then starts each one with the correct package manager. Live logs, ports and resource usage in one dashboard.
|
A full retrieval pipeline running entirely on your machine. Ollama for inference, LanceDB for vectors, and no API keys, because there is nothing to authenticate against.
|
|
An automation console for Windows. Run, schedule and monitor backup, sync and cleanup scripts from a single dashboard.
|
A README generator with a GitHub accurate live preview, eight templates, and an optional local model pass for prose. Runs fully offline.
|
The tier that decides whether the other two are fast. Schema first, indexes on purpose, and a query plan I have actually looked at.
|
Relational schema design under real access patterns. Viewers, creators and admins, with likes, comments, watchlists, uploads and server generated PDF analytics.
|
A browser extension and a web client sharing one SQLite file. That file is the entire backend. No accounts, no sync service.
|
| Concern | How I handle it |
|---|---|
| Source control | Git, feature branches, commits that explain why |
| Delivery | Docker, GitHub Actions, Vercel |
| Environment | Windows 11 with WSL2 |
| Design | Figma and Canva, enough to not need a designer for v1 |
| Currently studying | Advanced data structures, database internals, system design |
Two background jobs that never terminate. A manga backlog that grows faster than it drains, and a Valorant rank I have stopped defending in public.
There is a third. I tune and stress test my own machine the way other people tune a car, with fan curves, thermal thresholds and stability runs, none of it strictly necessary. Nothing has ever crashed. I simply want to know that it will not.
It turns out to be the same instinct that makes me read a query plan before shipping.
NEXT UP
├── Advanced data structures and algorithms
├── Database internals, including indexes, query plans and locking
├── Caching layers and background job queues
└── Production deployment and observability
EXPLORING
├── Offline first sync strategies
├── Embedded databases such as SQLite and LanceDB
├── Local model inference pipelines
└── Desktop packaging with Tauri and Electron
OPEN TO
├── Internships, backend or full stack
├── Open source developer tooling
└── Anything that removes a cloud dependency
