Projects
Machine learning work and working software. Screenshots and repositories where available.
Learnable spline activations as an alternative to MLPs
Built a KAN implementation exploring learnable spline-based activations, comparing behaviour against standard MLP layers.
Warehouse inventory desktop app, written as primary developer
A working OOP inventory system in Java and JavaFX with Maven: item and category CRUD, inbound and outbound stock transactions with audit history, low-stock alerts, and CSV report export through a centralized data service.
Full SDLC software engineering course project
Requirements analysis, context diagram, DFD, ERD, data dictionary, UI design, Java implementation, and black-box testing, delivered end to end with a team of three.
Four-model pipeline written up as an IEEE-format paper
A classification pipeline (SVM, Decision Tree, k-NN, Random Forest) with confusion-matrix evaluation; results written up in LaTeX using the IEEE format.
PKM-KC 2026 proposal for artifact classification
Co-authored a national student innovation program proposal for classifying Indonesian archaeological artifacts with a MultiTaskCNN Edge-Aware architecture.
Markdown editor
A beautiful online Markdown editor with live preview, built with Next.js.
Visualization of Maximum Clique Problem
An interactive visualization application for understanding the Maximum Clique Problem using the Bron-Kerbosch algorithm. Built with Python and Pygame, the application is designed for educational purposes, featuring a step-by-step walkthrough, narrative explanations, and a modern interface.
Built with vanilla JavaScript, HTML, and CSS. No framework, no build step, no dependencies.
A daily volleyball training app that runs entirely in the browser, works offline, and installs to your home screen like a native app. One routine, every day. The routine never changes, only the dial does.