Three projects from my degree, each with one honest number and my role.
Informal, code-switched Indonesian reviews often judge an aspect without naming it.
0.79
Sentiment macro-F1, IndoBERTBaselines: 0.35 to 0.54
Led the transformer models and evaluation. Co-annotated all 1,500 reviews.
Can dropout risk be flagged at enrollment, before first-semester grades exist?
0.838
Test-set recall, EarlyDrop appPrecision 0.537
Trained the app's model; charts and editing for the paper.
Measures a foot from one photo, using a bank card as a ruler.
Live demo · team of 5
I built the frontend and part of the backend around the team's computer-vision pipeline.