About
An independent software studio in Surabaya.
Aufal Tech is an independent software practice led by Naufal Hilmi Abdurrahman, building web, Android and desktop applications from Indonesia.
Vision
Software should be judged by the problem it removes.
A product matters when someone’s work gets shorter, clearer or possible at all because of it. That is the measure used here, ahead of feature counts and ahead of which framework is fashionable.
It is why the portfolio is small and specific. Each project starts from a concrete situation: a learner with no signal, a job seeker comparing five boards, a customer trying to picture a rented desk.
Engineering approach
Understand first. Then build the smallest thing that is right.
Work begins by reading the problem end to end before choosing a solution. The first version is deliberately narrow, and the scope of what it leaves out is written down next to what it includes.
Difficult parts are isolated instead of spread around. In Laras the one component that needs native libraries runs in its own process, so the rest of the backend stays simple to build and test.
Development standards
What every project is held to.
Layered code
Business rules are separate from HTTP, UI and storage, so they can be tested without starting a server or opening a window.
Tests where logic lives
Branches, parsers and scoring functions get tests. Trivial glue does not get ceremony.
Written decisions
Each project carries an architecture note explaining what was chosen and why, in the repository next to the code.
Repeatable releases
Builds come from a pipeline triggered by a version tag, reviewed as a draft before they are published.
Honest interfaces
Missing data, limited coverage and unfinished features are shown to the user instead of hidden behind a default.
Privacy by placement
Sensitive data is processed where it already is, on the device or on a server you run, whenever the task allows it.
Product philosophy
Boring where possible. Careful where it counts.
Most of a product should be unsurprising: standard libraries, platform conventions, a database that is one file when one file is enough. That leaves attention for the part that is actually hard.
Rules come before models. A transparent baseline ships first and stays as the fallback, and machine learning is added at the single step where it clearly does better.
A prototype is called a prototype. This site labels what is released, what is live, what is source only and what is still an early build, because trust depends on that distinction.
Have a problem software could solve?
Describe it in a few sentences on WhatsApp. You get a direct reply about scope, platform and a sensible first version.
Request a Project