Taimoor Safdar

I've built and deployed machine learning applications, including a deepfake human-face detector that achieved 86% accuracy on a balanced 40,000-image dataset. I improved its baseline from 78% through data augmentation and hyperparameter tuning, then containerised it with Docker and deployed it on Streamlit Cloud.

At NUST's Optical Networks and Technologies Lab, I configured BGP routing on SONiC switches and wrote Python automation for interface provisioning. I also researched Zero Trust Architecture for SONiC, including its advantages and trade-offs.

During my information security internship at Cybersecurity Zone, I helped develop and test automated modules for the NVAS vulnerability scanner using Nmap, Metasploit, CVE analysis, Python, and a pre-trained AI model. In a testing environment, I scanned 100 hosts and identified three CVEs.

I'm also building Glide, a peer-to-peer carpooling platform with verified drivers, real-time GPS tracking, smart ride matching, dynamic pricing, and SOS alerts. Alongside ML and security work, I've built a Python/Pygame chess bot and collaborated on a responsive sustainable-fashion e-commerce website.