Michael Xu

Lead Ai Engineer @Beroai

Montreal, QC, CA

Led a team of two AI engineers and two data specialists to launch the company’s first public demo- Built an end-to-end ML pipeline by integrating teammates’ models to process real-time sensor data- Supervised EDA across modalities and reworked the data collection protocol, reducing synchronization errors from 3 seconds to 200 milliseconds- Prototyped models including keyword spotting, IMU prediction, and emotion classification, achieving 70% accuracy on unseen scenarios- Reviewed state-of-the-art on-device LLMs (e.g, Phi-4, Gemma-3) and alignment strategies (e.g, RLHF, DPO, GRPO) to guide the team’s next phase of ML development

University of San Francisco

Bachelor of Science - BS, Data Science

McGill University

Master of Science - MS, Computer Science