Published research, ML pipelines, and data-driven decisions — not just portfolio pieces.
Frontend Development · Live Product
Invoice Generator | QRIS-enabled (Personal Freelance Toolkit)
A personal invoicing tool I built to streamline my freelance workflow, from creating branded invoices to receiving instant payments via my own QRIS.
Designed to eliminate repetitive admin work and speed up payment collection without relying on third-party platforms.
Vanilla JS
HTML5
CSS3
jsPDF
✦ Deployed to Netlify · 100% Client-Side Architecture
Machine Learning · Published Research · Thesis Research
ML Algorithm Comparison for Cosmetic Sales Prediction
Compared 4 classifiers (Random Forest, XGBoost, Logistic Regression, Naive Bayes) on Tokopedia data. Used SMOTE for class imbalance and GridSearchCV for hyperparameter tuning, identifying Random Forest as the superior model.
Random Forest
XGBoost
SMOTE
GridSearchCV
✦ 85.3% accuracy · F1-Score 0.75 · JURTEKSI, Sept 2025
Data Analytics · Published Research
LPG 3KG Distribution Clustering
Applied K-Means clustering with Elbow Method to analyze LPG distribution equity across sub-districts in Palembang. Revealed suboptimal outlet ratios in dense areas like Sukarami, enabling better logistics planning for subsidy distribution.
Python
Pandas
Scikit-learn
Matplotlib
✦ Published · Jurnal Buffer Informatika, April 2025
Data Analytics · Published Research
K-Means Clustering for F1 Driver Performance (2024)
Segmented the 2024 F1 grid into 4 performance tiers using K-Means + Elbow Method + Silhouette Scoring (0.5735). Revealed hidden patterns between driver consistency and peak lap speed, providing strategic insights for racing teams.
Python
K-Means
Scikit-learn
Lead Researcher
✦ Published · Jurnal Buffer Informatika, Oct 2025
Data Visualization · Streamlit Dashboard
F1 Performance Analytics Dashboard
Bridging the gap between raw Python analytics and actionable insights. I translated my K-Means clustering research into a high-fidelity interactive dashboard, focusing on driver consistency and performance tiers. A complete study on how complex ML data should be presented to stakeholders.
Python
K-Means
Scikit-learn
Lead Researcher
✦ Published · Jurnal Buffer Informatika, Oct 2025
Machine Learning · Bangkit Capstone
Sisain — Food Waste Reduction App
Led data preprocessing and CNN model development to classify food waste types. Managed documentation, product-market fit strategy, and kept a cross-functional team aligned from ideation to final submission.
Python
TensorFlow
CNN
Project Manager
✦ Sustainable Futures theme
