Article

I’ve been working on a lightweight computer vision app that recognizes PPE in images and videos. Upload an image or video, and it returns annotated outputs plus the detected class labels (e.g., helmet, vest, gloves). Perfect for quick demos, audits, or integrating into ops dashboards. What it does:Detects 6 PPE classes in images and uploaded videosShows annotated frames and a … Read more

Excited to share my latest project: Building a Computer Vision Model with YOLOv8 to Count Pipes! 🚀 As a machine learning enthusiast, I recently dove into object detection to solve a real-world problem—accurately counting pipes in industrial settings. Using YOLOv8, I trained a custom model on a dataset sourced from Google Images, and the results … Read more

Migrating your homelab reverse proxy from Nginx to Caddy is way easier than it sounds—especially when you do it without downtime. Below is a straight-to-the-point guide you can copy-paste into WordPress. No Markdown, just clear steps and bullet points. What you’ll get out of this post – Full backup of every Nginx config and TLS … Read more

In a world where artificial intelligence (AI) is rapidly weaving itself into the fabric of everyday life—from healthcare decisions to creative endeavors—a groundbreaking white paper titled Doing AI Differently calls for a paradigm shift. Released by The Alan Turing Institute, the University of Edinburgh, and the UK’s Arts & Humanities Research Council (AHRC-UKRI), this document … Read more

Artificial Intelligence has the potential to revolutionize education — but only if it’s designed with the right principles in mind. An effective AI learning companion isn’t just a “question-answer machine.” It should guide, challenge, and inspire learners while also supporting educators. Here’s what it takes to make that happen. 1. Personalized Tutoring and Guidance The … Read more

In the fast-evolving world of large language models (LLMs), the Mixture-of-Experts (MoE) architecture has emerged as a game-changer—offering high performance without the massive computational cost of fully dense models. Two recent heavyweights—Alibaba’s Qwen3 30B-A3B and OpenAI’s GPT-OSS 20B—are pushing the boundaries of what MoE models can do. Released in 2025, these models represent two distinct … Read more