Projects

Developed a digital twin of Topdrive and Drawworks drilling systems in MATLAB/Simulink and Simscape. Generated Beckhoff Structured Text from the model and validated PLC control logic via hardware-in-the-loop (HIL) testing.

Self-driving strategy using a ResNet-18 CNN for lane following. Designed a digital PID controller with pidTuner and compared stability and robustness across controller configurations.

Designed and simulated EMK, Adaptive, and Robust (Sliding-Mode) controllers for a 2-link robot in Simulink. Evaluated stability, control effort, and robustness across all three approaches.

Classified terrain types from IMU sensor data (accelerometer + gyroscope) using Conv1D, LSTM, GRU, and dense layers in Keras, achieving 85% accuracy with the best RNN LSTM model.

Deep learning pipeline for animated character face recognition achieving 93.49% accuracy on a real-time dataset.

Neural network trained via backpropagation from scratch using NumPy on the MNIST handwritten digit dataset (no deep learning frameworks used).

Embedded system in C for real-time energy monitoring, usage optimization, and equipment condition monitoring.