Rashid
Samid detects drones and identifies their type from sound. I trained and fine-tuned the audio transformer, worked on data augmentation, and built the website that shows the results.
The project was selected from a pool of over 13,000 students, presented to the governor, and later incubated by GADD.
Company knowledge gets scattered across files, and finding one document is only part of the problem. I built Misbar to search connected sources, trace how people and projects relate, and flag conflicting documents. It keeps answers linked to their sources and respects file permissions.
I built the full application, from document processing to the interface. Misbar won first place at Buildathon, organized by Riyadh Valley.
AceWallet is a wallet prototype that helps you pick the card with the best cashback or rewards for what you are buying. It was selected for Tuwaiq Academy's Amad Hackathon from thousands of applicants.
Farras is a project on detecting AI-generated Arabic text. In our first study, we compared character n-grams, linguistic features, and AraBERT. Combining the first two reached 93.16% accuracy in our internal evaluation and 86% on a separate set of 348 samples.
