Xiaochi Cai

Offering solid foundation in research principles and keen interest in developing within research environment. Contributes strong analytical mindset and ability to quickly learn and apply new research techniques. Ready to use and develop data analysis and literature review skills in M.S and Ph.d program.

LLMs From Scratch — Transformer Implementation

Reproduced core components of a transformer-based language model following Rasbt’s open-source framework. Implemented tokenization, attention mechanisms, positional encodings, training loop, and evaluation scripts. Extended the base implementation with personal experiments (e.g., alternative embedding variants, ablation on training configurations).

Deep Reinforcement Learning Implementations — Hugging Face Deep RL Course

Completed the practical modules of the Hugging Face Deep RL Course by implementing multiple RL agents with Stable-Baselines3 and Gym environments. Trained and evaluated agents (e.g., CartPole, LunarLander) and conducted experiments to compare algorithm behaviors (DQN, PPO, policy-gradient methods). Uploaded trained models and reports to the Hugging Face Hub as part of the workflow.

• International Student Investment Association (ISIA) – Core Member & Weekly Presenter
• Delivered weekly presentations on global market trends, enhancing analytical and public-speaking skills.
• Published summarized insights on social media platform to improve writing and audience engagement.