About the studio
A studio built to ship, not to staff up.
Latellu designs, builds, and hands over production software on a fixed weekly cadence. The same process runs Latellu's own products: Atlas CMS serves this site, Tumbuhku tracks child growth against WHO and IDAI standards at app.tumbuhku.id, and Pulse has been publishing in English, Indonesian, and Japanese since March 2026.
Every engagement ends with you holding the codebase, the tests, and the deployment pipeline. You have owned them since the first commit, because the repository lived in your accounts from day one.
The engagement
One team, from kickoff to handover.
01
A fixed price before work starts
Scope is written down and priced before the first commit. The number you see is the number you pay; if the estimate was wrong, that cost is ours. Nothing is billed by the hour, so slow work is never in our interest.
02
Working software every week
Every week ends with something you can click through in your own repository. If it is wrong, saying so costs you one week, not one quarter.
03
Handover from day one
Repository, credentials, and infrastructure live in your accounts from day one. The last day of the engagement changes nothing, because there is nothing left to hand over.
04
After, on your terms
Continued work is its own fixed scope. Or you take the codebase to your own team, which was the plan from the start.
How the work actually runs.
You talk to the person writing the code.
The person in the kickoff call is a person in the commit history, so questions get answered from memory, not a ticket queue. The team grows only when a project needs more hands.
That also means bad news arrives fast. If something is slipping or an approach is not working, you hear it the week it happens, from the person who knows why.
Decisions are written down.
Every non-obvious choice gets a short note in the repository: what was decided, what was rejected, and why. Commit messages say why, not just what.
This is what makes the handover real. The next engineer inherits the reasoning, not just the code, and never has to schedule a call with us to understand their own system.
Boring choices, deliberately.
The stack below is proven, documented, and hireable. Every tool in it is picked so that the next engineer already knows it, the failure modes are on record, and nothing needs us around to keep running.
When a project genuinely needs something unusual, it gets it, written down with the reason. Novelty has to be argued for; the boring option is the default.
Built with
- TypeScript
- React
- Next.js
- Go
- Node.js
- PostgreSQL
- Tailwind CSS
- Docker
- AWS
- GitHub Actions
Think we are a fit?
Tell us what you are building. We will come back on whether we are a fit, roughly how long it would take, and how the work would run.