Blog

Thoughts on building production AI systems — from idea to deployment.

·2 minmachine-learningcomputer-visionpython

I Replaced a 730 MB Cartoon Model with a 78 MB One — and It Got Better

How I went from a 730 MB pix2pixHD model that memorised its training data to a 78 MB U-Net that actually generalises — using FLUX as a teacher, a face refiner for small faces, and a lot of failed experiments.

·6 minyolocomputer-visionpython

How I Solved YOLO Bounding Box False Positives with a Dual-Model Detection Strategy

No single YOLO model gets bounding boxes right for every photo. Here's how running two models in parallel and merging their detections intelligently eliminated both false positives and missed detections in our production background removal service.

·2 minaicomputer-visioncareer

Why I Build AI Systems That Ship, Not Just Demo

Most AI projects die between prototype and production. Here's how I bridge that gap — from theme park photo engines to LCL freight tools.

© 2026 thearifoffice · Made with careBangladesh ↔ GCC · Singapore · Malaysia · Worldwide