- Working at the Data Systems Group on an LLM Agent for data annotation.

- Applied reinforcement learning to optimize the project antenna's physical characteristics, utilizing the Q-Learning algorithm. Initially trained the agent with random exploration across the action space, progressively guiding it towards optimal actions by converging exploration to exploitation based on predicted future rewards.- Developed and refined reward mechanisms, employing a train-validate-test methodology to fine-tune key hyperparameters such as learning rate, discount factor, and exploration rate.- Implemented SOLID principles to ensure modularity, maintainability, and code style consistency throughout the entire development process.- The improved script reduces each training loop episode by 15 seconds, cutting the overall simulation runtime by 33%.

Business intelligence, database management, etc.