AI & Machine Learning
I specialize in applying deep learning to real-world problems, with a focus on medical imaging, computer vision, and simulation prediction. These projects showcase my ability to design and deploy models using TensorFlow, OpenCV, and custom CNN architectures.
🧠 Glioma Detection with U-Net
Goal: Segment brain tumors across multiple MRI sequences using a U-Net model.
- Preprocessed multi-modal MRI data (T1, T2, FLAIR, T1CE)
- Created RGB-encoded segmentation maps by merging modalities
- Trained a U-Net CNN for tumor boundary localization
- Balanced model complexity to address underfitting/overfitting
🔁 Predicting John Conway’s Game of Life
Goal: Use deep learning to predict the future state of a cellular automaton.
- Compared custom CNN vs VGG-16 on simulated life evolution data
- Converted 2D grids into 4D tensors to match model input
- Trained both binary and multi-class output models
- Achieved strong F1-score and recall on unseen patterns
🧪 Key Skills Demonstrated
- Deep Learning: U-Net, VGG-16, custom CNNs
- Medical Imaging: MRI segmentation, modality fusion
- Model Evaluation: F1-score, precision, recall, loss tuning
- Python Libraries: TensorFlow, OpenCV, NumPy
- Data Engineering: Preprocessing, normalization, 4D input shaping