emoji_events
Letter of CommendationJune 2026
NUS SCALE · AI & Machine Learning (Deep Learning)
Weapons Image Classification — Capstone Project
97.2% accuracy on the original dataset
The capstone of a four-day NUS SCALE Deep Learning course, judged on the best model across the cohort. The project went past a single accuracy number: it audited the provided dataset, found 31% duplicate leakage inflating results, rebuilt a leakage-proof split, and reported honest before/after numbers across four models, capped with a live interpretability demo. The commendation recognized the depth of process, not the accuracy alone.
Read the case studytrending_flat