Aubrey Lubasi

Aubrey
Lubasi

I build tools that check whether something can be trusted: a payment, a message, a news feed.

Final-year Bachelor of Cyber Security (BCS) student at Zambia University College of Technology (ZUT) in Ndola, focused on network security.

This is the SHA-256 fingerprint of my name. It doesn’t contain the text and can’t be turned back into it.

Projects

VoucherGuard ZM

Final-year project

Stops payment fraud in accounts offices. Every payment voucher is digitally signed by the person who prepares it and the person who approves it. If anyone changes a beneficiary's account number or an amount afterwards, payment is blocked and the exact change is shown. Every action goes into a tamper-evident audit trail.

Built with Node.js, React, SQLite and Ed25519 signatures, with an Android and iPhone app.

ZSL Translator

In progress

Turns Zambian Sign Language into text and speech with just a phone camera, and works with no internet. It learned the full ZSL alphabet from one teaching video by reading the letter shown on screen, so nothing had to be labelled by hand. It keeps only hand positions and never saves video, so the people who sign for it stay anonymous.

Built with Python, MediaPipe and scikit-learn, with an offline Android app.

Zed Headlines bot

Live

I built this because I hated the ads and unrelated posts on the Zambian news pages I followed. Zed Headlines posts Zambian news and nothing else, around the clock. Every 20 minutes the bot checks seven news sites, skips stories already posted by another site, filters out sponsored content and limits how often it posts. It runs free on GitHub Actions, so no computer has to stay on.

Built with Python, GitHub Actions and the Facebook Graph API.

ZM Phish Hunter

In progress

Finds new websites that imitate Zambian banks, mobile money services and public bodies, before the scam link starts spreading. Every new HTTPS website is recorded in public certificate logs. The bot searches those logs for names that look like a Zambian brand, skips the real sites and scores how suspicious each one is. It is passive: it never visits or scans a site. Still in progress: the scoring works, and screenshots and page comparison are next.

Built with Python, Certificate Transparency logs and GitHub Actions.

ScamShield

In progress

Helps people spot scam messages before they reply or send money. I'm currently testing it on WhatsApp.

Telegram study and job assistant

Personal tool

A private Telegram bot that turns my lecture notes into flashcards and searches for internships and jobs for me.

Built with Google Apps Script and the Telegram Bot API.

What I work with

Security

  • Web app testing with Burp Suite
  • Network scanning with Nmap and Nuclei
  • CTF challenges, forensics and steganography
  • Malicious document analysis and YARA rules
  • Network design in Cisco Packet Tracer

Building software

  • Node.js and Express
  • React
  • Laravel and PHP
  • Python
  • Computer vision and machine learning with MediaPipe and scikit-learn
  • Android, Flutter and Capacitor apps
  • Google Apps Script

Data and infrastructure

  • SQLite and PostgreSQL
  • Docker
  • GitHub Actions
  • Linux

Contact

I'm looking for internships and graduate roles in network security and software development.