Projects

Deep learning model for predicting future video frames using temporal convolutions and recurrent architectures.

Implementation of Kalman filter for multi-sensor data fusion, combining noisy sensor readings into accurate state estimates.

Real-time face detection and tracking pipeline using classical CV techniques for robust performance.

Object detection combined with tracking algorithms for persistent identification across video frames.

GPU programming exercises exploring parallel computing with CUDA — kernels, memory management, and optimization.

Assignments from the DLVR course at Uni Bonn covering CNNs, object recognition, and image classification.

Classic image classification pipeline using bag-of-visual-words with SIFT features and SVM classifiers.

MIT xv6 OS labs — implementing system calls, scheduling, virtual memory, and file systems from scratch.