Dashboard SAKTI

The problem

Otorita IKN has six budget units, called satker, and each one's budget and spending data lives in SAKTI, the Ministry of Finance's system. Whenever the planning bureau needed figures for a report, an executive review or Power BI, someone logged into SAKTI, downloaded an Excel file for each unit, and stitched them together by hand.

That meant no history and no single view. Nobody could see how spending moved over time. Every other tool that needed the numbers got them by copying and pasting. The data team wanted the figures collected automatically, kept over time, and shown in one place.

What I built

A nightly pipeline that fetches every unit

A scheduled job runs every night at 02:00 WIB. It pulls six kinds of SAKTI data for every active unit, in parallel, and saves them to a database. The list of units comes from the database when the job starts. An administrator adds or switches off a unit in the admin panel, and the next run follows. Nothing about the units is hardcoded. Every run is safe to repeat, because saving the same record twice updates it and never duplicates it.

I simulated SAKTI so work never waited on access

Real SAKTI credentials weren't available yet. SAKTI's login is unusual: every response hands back a new one-time key, and the previous one stops working at once. I built a small stand-in server that behaves the same way, including expired-key responses, and serves sample data for all six data types. The whole pipeline runs against it. Moving to the real system means changing two settings, with no code changes.

One place for the save logic

The scheduled jobs write to the database through the same data models the web app uses. I chose this over putting a web API between them, so the rules for saving a record live in one place. The cost is that the scheduler depends on the web app's code. At six units, that trade is fine.

A dashboard with a sync badge and Excel export

The dashboard has three pages. The overview shows total budget, total spending and the share absorbed, with a trend chart and a gauge. A detail page breaks one unit down by program and activity, with date filters. A management page lists the units. Each nightly run records its status, row count, duration and any error, and the dashboard shows when data was last synced. Any view can be exported to a formatted Excel file.

Result

I built the platform in two days in June 2026, from an empty repository. It works end to end: nightly job, SAKTI client, database, API, dashboard and Excel export. Today it runs on the simulated SAKTI server with sample data. The production setup is ready and tested to start cleanly. Real data waits on the Ministry of Finance providing credentials. The dashboard has no login of its own in this first version, and is meant for OIKN's internal network.