Recent Software Engineering graduate from FAST-NUCES with a focus on high-impact AI systems. Developed GeoShield, an AI disaster monitoring system that automates damage assessment and authority notification within seconds. Experienced in building end-to-end ML pipelines with TensorFlow and PyTorch, achieving significant model performance gains (35% mIoU improvement). Passionate about leveraging technology for social good and disaster response.
Islamabad, Pakistan
Experience
Dec 2025 – Apr 2026
WORK
Machine Learning Intern
KP IT Board
Peshawar, Pakistan
Built an end-to-end image preprocessing pipeline for Computer Vision and Machine Learning tasks. Cleaned and preprocessed image datasets to improve data quality and model performance. Applied Gaussian filtering for noise reduction and Bilateral Histogram Equalization for image enhancement. Extracted image features using the Canny Edge Detection algorithm. Performed data optimization techniques, including dataset balancing, normalization, and batch processing to improve training efficiency and model robustness. Optimized data workflows to support scalable and efficient AI model development.
•Built an end-to-end image preprocessing pipeline for Computer Vision and Machine Learning tasks.
•Cleaned and preprocessed image datasets to improve data quality and model performance.
•Applied Gaussian filtering for noise reduction and Bilateral Histogram Equalization for image enhancement.
•Extracted image features using the Canny Edge Detection algorithm.
•Performed data optimization techniques, including dataset balancing, normalization, and batch processing to improve training efficiency and model robustness.
•Optimized data workflows to support scalable and efficient AI model development.
Key Achievements
→Improved mIoU from 0.339 to 0.5469 (35% increase) through optimized preprocessing pipeline.
→Managed and preprocessed a dataset of 10,000 images (5,000 paired sets) to enhance model training efficiency.
Developed a machine learning model using Linear Regression to predict student admissions. Implemented with Python and Gradio for an interactive interface.
AI-based disaster monitoring system designed to bridge the gap between manpower and disaster zones by providing near-instant analysis and automated alerts to authorities.