I am a Computer Engineering undergraduate interested in building reliable and robust machine-learning systems. My current work focuses on adversarial robustness, especially how models trained against one threat model behave under attacks defined by another norm.

I enjoy taking projects through the full experimental cycle: implementation, controlled evaluation, error analysis, visualization, and reproducible documentation.

• Adversarial attacks and adversarial training
• Cross-norm robustness under L∞ and L2 threat models
• Reliable evaluation of machine-learning systems
• Deep learning and computer vision
• Mathematical foundations of machine learning

An empirical study of adversarial robustness across L∞ and L2 threat models on MNIST and CIFAR-10. The project implements FGSM/PGD-style evaluation, adversarial training, epsilon sweeps, robustness matrices, and reproducible experiment outputs.

An end-to-end 37-class image-classification pipeline using ResNet18 and PyTorch. I compared frozen-backbone and partial fine-tuning strategies and evaluated the final model with per-class metrics and confusion-matrix analysis. The fine-tuned model reached 87.84% test accuracy and 87.69% macro F1.

A sequence-labeling pipeline for Persian NER, including preprocessing, batching, model training, evaluation, and overfitting checks. The best validation F1 recorded during development was 0.8952.

Implementations of classical AI and machine-learning methods, including A* search, Minimax with alpha-beta pruning, constraint satisfaction, Hidden Markov Models, Decision Trees, Random Forests, Value and Policy Iteration, Q-Learning, and SARSA.

• Victim Cache Simulator — memory-hierarchy simulation in Go using Akita
• Computer Architecture — processor datapaths, pipelining, caches, and Verilog modules
• CollarAI — a Chrome extension for real-time body-language feedback during video calls

Languages: Python, C++, Java, Go, Verilog
ML and data: PyTorch, NumPy, pandas, scikit-learn, Matplotlib, Jupyter
Tools: Git, Linux, Google Colab

• Linear Algebra — vector spaces, eigenvalues, SVD, PCA, matrix calculus, and optimization
• Probability and Statistics — probability theory, random variables, inference, and ML foundations

I am interested in research opportunities related to adversarial machine learning, robustness, and reliable deep learning.

Email: moradi.mahdi.ce@gmail.com