XG-Guard Receives the Merit Award at the Bangladesh ICT & Innovation Awards 2026

I am honored to share that XG-Guard received the Merit Award at the Bangladesh ICT & Innovation Awards 2026, in the Student (Tertiary) - Business Services category.

This is a big moment for me, and I want to use this post to explain what XG-Guard is, why I built it, and what the results look like so far.

The problem: many Bangladeshi websites are not safe

Cyber attacks are not only a problem for big companies. Small businesses, schools, news sites and online shops are attacked every day. In Bangladesh, an estimated 60%-85% of active websites contain serious vulnerabilities. That means attackers can often steal data, change the content of a page, or inject harmful scripts.

There are good security tools on the market, but they have a cost. Foreign web application firewalls are expensive, and many local developers and small business owners simply cannot pay for them. As a result, thousands of websites stay unprotected.

What is XG-Guard?

XG-Guard is an AI-powered web application firewall (WAF) built to protect Bangladeshi websites from cyber attacks in real time.

A WAF sits between the internet and a website. Every request that comes to the site passes through it first. A good WAF can tell the difference between a normal visitor and an attacker, and it stops the attacker before anything reaches the server.

XG-Guard works as a reverse proxy. It analyzes every incoming web request and uses an AI classifier to find harmful payloads. If a request looks like an attack, XG-Guard blocks it. Because it is a reverse proxy, it is compatible with any website.

Which attacks does it stop?

XG-Guard is designed to block some of the most common and most damaging web attacks:

• SQL injection (SQLi): the attacker sends database commands through a form or URL to read or change private data.
• Cross-site scripting (XSS): the attacker injects a harmful script into a page that other users will open.
• Command injection: the attacker tries to run commands directly on the server.

These attacks can lead to data leaks, website defacement and script injection, which is exactly what XG-Guard helps to prevent.

How it is built

The firewall engine is written in Go, which is known for handling many requests quickly. The AI classifier was built with Python and Keras. Together, they make a system that can decide whether a request is safe or harmful without slowing down the website.

Results from testing

I tested XG-Guard in a controlled lab environment. These are the main results:

• 99.93% detection accuracy in testing.
• 35% lower latency compared to traditional WAFs.
• 10,000+ simulated attack payloads blocked in lab tests.

These numbers come from lab tests with simulated attacks, so real-world traffic may behave differently. Still, they show that an AI-based approach can be both accurate and fast.

Built to be affordable and easy to understand

Price was one of the main reasons I started this project. XG-Guard can be deployed for less than $40 per year, which makes professional-level protection possible for small websites and local developers.

I also wanted the tool to be useful for people who are not security experts. XG-Guard gives simple explanations of the attacks it blocks, so a website owner without a technical background can still understand what happened.

Looking ahead: Bangla-language threat data

XG-Guard is designed to support future Bangla-language threat datasets. Most security research uses English data, but attacks on local websites can include local language and local patterns. Supporting Bangla datasets is a step toward security tools that understand our own context.

What this award means

For me, the Merit Award is a sign that protecting local websites with local AI research is a topic worth working on. My hope is that XG-Guard helps improve national cyber defense, encourages more local AI research, and inspires other students to build practical tools for real problems in Bangladesh.

If you want to see the technical details, you can read the full XG-Guard project page. Thank you for reading!