Models, mechanisms
& algorithms

Gas-turbine regression

I used five years of sensor data to predict total energy yield and emissions, including CO and NOx. I selected features using correlation and compared linear, polynomial, and decision-tree models using RMSE and R-squared.

Text-prompted segmentation

For a separate project, I combined TF-IDF and logistic-regression intent classification with GroundingDINO detection, SAM mask generation, and mask refinement. I encoded the masks in run-length format for evaluation.

The segmentation project placed first of 75 on the class private leaderboard.

Kinematics and trajectories

I modeled a three-degree-of-freedom chess arm in MATLAB using Denavit-Hartenberg parameters. I implemented forward and inverse kinematics, generated trajectories between target positions, and checked the motion in 3D simulation.

Ultrasonic sensing and navigation

I built an Arduino-based maze-solving robot with ultrasonic sensing. Its navigation code uses range measurements for wall following, obstacle detection, collision avoidance, and turn decisions during autonomous maze traversal.