Nexux Labs

Technology

The stack we build on, and why each piece is in it

A technology list on its own tells you nothing. Here is what we use, what it is for, and the reasoning behind each choice.

How we choose

Four rules behind every technical decision

These are the rules we apply before adding anything to a project, including when a newer option looks more interesting.

01

Boring beats clever

We pick tools with large communities and long support windows. If you need to hire a developer in two years, you should be able to find one easily.

02

One language where possible

TypeScript across web, mobile and server keeps a small team effective and makes shared logic genuinely shared.

03

The hosting you have counts

A stack that needs infrastructure you cannot afford to run is the wrong stack. We build to what you actually host on.

04

No lock-in by design

Standard frameworks, your own repository, your own accounts. Replacing us should be inconvenient, never impossible.

The stack

Everything we use, grouped by what it is for

Front end

What your users touch. Chosen for render speed, accessibility and being maintainable by any React developer you hire later.

  • Next.js

    Server rendering, static export and routing in one framework

  • React

    The component model the rest of the stack assumes

  • TypeScript

    Catches the class of bug that reaches production otherwise

  • Tailwind CSS

    Consistent spacing and colour without a growing stylesheet

  • Framer Motion

    Motion that respects reduced-motion preferences

  • Storybook

    Components reviewed in isolation before they reach a page

Back end

APIs, business logic and integrations. Boring, typed and testable beats clever every time.

  • Node.js

    One language across front and back end for small teams

  • NestJS & Express

    Structure for anything beyond a handful of endpoints

  • Laravel

    When the project lives on PHP hosting and should stay there

  • Python & FastAPI

    Data, scripting and anything touching a model

  • REST & GraphQL

    Chosen per project, documented either way

  • Stripe, eSewa & Khalti

    Payments including the failure paths people skip

Mobile

Cross-platform first, so you ship to both stores from one codebase and one budget.

  • React Native

    Shares language and people with the web build

  • Flutter

    Strong choice when the UI is heavily custom

  • Expo

    Faster builds and over-the-air fixes

  • Firebase

    Auth, push and crash reporting without running servers

  • SQLite & MMKV

    Local storage so the app survives a dropped connection

Data

Where the real performance problems live, and where most of them get solved.

  • PostgreSQL

    Our default: strict, fast and good with JSON

  • MySQL & MariaDB

    For hosting and products built around it

  • MongoDB

    Where documents genuinely fit better than rows

  • Redis

    Caching, queues and rate limiting

  • Prisma & Drizzle

    Typed queries and migrations under version control

Cloud & DevOps

Deployment, monitoring and backups. The layer that decides whether a release is routine or an event.

  • Docker

    The same environment locally, on staging and in production

  • GitHub Actions

    Tests and deploys run from the repository, not a laptop

  • AWS & DigitalOcean

    Sized to real traffic rather than a diagram

  • Nginx & LiteSpeed

    Caching and TLS done properly

  • Cloudflare

    DNS, CDN and a WAF in front of the origin

  • cPanel

    Because plenty of good Nepali hosting is cPanel, and that is fine

Quality & analytics

How we know something works, and how you know it is being used.

  • Playwright

    End-to-end tests on the journeys that earn money

  • Vitest & Jest

    Unit tests where logic is worth protecting

  • Lighthouse & Core Web Vitals

    Performance measured, not asserted

  • GA4 & Search Console

    Traffic and query data you can act on

  • Sentry

    Real errors from real users, with a stack trace

  • Uptime monitoring

    You hear it from us before a customer

Straight answer

What we do not claim to be

Picking a partner is easier when you know the edges of what they do. We would rather tell you now than three weeks into a project.

We are not a blockchain shop

We will build the web and mobile layer around a chain integration, but if your project is primarily protocol engineering, you want specialists.

We do not train large models

We integrate existing models and build retrieval and workflow around them. Training a foundation model is a different business with a different budget.

We are not a staffing agency

You get a small team that works on your project, not a stream of CVs to interview.

We do not do one-day websites

If you need something live tomorrow, a good template on a hosted builder is genuinely the better answer, and we will tell you that.

More answers on the FAQ page

Not sure which stack your project needs?

Describe what you are building and what you can host. We will recommend a stack and tell you what it means for cost and hiring later.

whatsapp-icon Nexux