Handwritten Digit Recognition — Developed a Convolutional Neural Network (CNN) to classify handwritten digits from the MNIST dataset. The model achieved 97% accuracy, leveraging multiple convolutional and pooling layers, ReLU activations, and dropout regularization to prevent overfitting. Implemented using TensorFlow/Keras, the system can efficiently recognize digits in real-time and demonstrates practical applications in automated form processing and digit recognition tasks.
Object Recognition and PID Line-Follower Robot (Webots Simulation) — Designed and programmed an autonomous robot in Webots, a virtual simulation platform, to follow a designated path using PID (Proportional-Integral-Derivative) control for precise navigation. Integrated computer vision techniques for real-time path tracking and obstacle detection. The simulation allowed testing and optimization of control algorithms and vision-based decision-making without requiring physical hardware, demonstrating skills in robotics, control systems, and algorithmic problem-solving.
Network Design (Cisco Packet Tracer) — Simulated a scalable network infrastructure for a multi-department institution using Cisco Packet Tracer. Configured routers, switches, VLANs, and routing protocols (such as OSPF and EIGRP) to ensure optimized communication, security, and redundancy across departments. The project showcased practical skills in network planning, troubleshooting, and performance optimization in an enterprise environment.
Sound and Air Pollution Monitoring System (IoT) — IoT-based Air and Sound Pollution Monitoring System for BRACU Street that continuously measures harmful gases and noise levels using sensors connected to an Arduino. Data is transmitted via Wi-Fi to an Android app, allowing authorities and citizens to monitor pollution in real-time and take action to maintain a safe environment .
Sudoku Solver — Implemented a Sudoku solver following the Model-View-Controller (MVC) pattern, which separates the user interface, data management, and algorithm logic. Used backtracking algorithms with optimization heuristics to solve puzzles efficiently. This project highlights algorithmic problem-solving, software design principles, and efficient computational methods for combinatorial problems.