Hamza Azeem

I’m a Data Science researcher focused on AI/ML pipelines and deep learning for computer vision. I build and evaluate end-to-end models in Python, with an emphasis on performance, reliability, and GPU-based training.

In my cross-domain deforestation detection work, I led a system that outperformed a published baseline by +10.2% to +17.6% F1-score. I combined domain adaptation with uncertainty-aware prediction, and I validated results across multiple domain pairs using ablation studies and automated reporting.

My practical toolkit spans data preprocessing, model training, and rigorous performance evaluation for ML/DL workflows. I also enjoy building from scratch—like implementing Canny edge detection in Python, including Gaussian smoothing, gradient computation, non-maximum suppression, and hysteresis thresholding.

I’m currently completing my BS in Data Science (GIFT University) and I’m eager to apply my research mindset to real-world problems. I’m drawn to teams that value experimentation, clean pipelines, and measurable impact in AI/ML and computer vision.