Jon Xu

Worked as an AI engineer in a team of five. Wrote OCR postprocessing code improving accuracy from ~75% to >99%. Developed inference and deployment pipelines using Triton Inference Server. Load tested with Locust, making adjustments to match specifications. Implementation of YOLOv7 into Kubeflow, providing useful performance/results tracking for the team, as well as significantly improving over previous best results on small object detection. Development of an object tracking application based on drone footage and flight records for a construction company: responsibilities included experiment management, development tasks, feedback, requests and communication with the client (in Japanese). Development of novel ML-based workflows in Temporal incorporating the ChatGPT OpenAI API. Development of Retrieval Augmented Generation using LlamaIndex, enabling 'smart' queries to be performed on company documents.

Postdoctoral research in algebraic combinatorics, representation theory and graph theory.

I was a tutor for many mathematics courses ranging from first year differential calculus to final year abstract algebra. This experience taught me how to speak in front of audiences, and how to communicate highly technical concepts in a simple and clear manner.

PhD (Mathematics). My PhD thesis was in representation theory, an area of pure mathematics.

Bachelor of Science (Advanced Mathematics). Graduated with First Class Honours. Developed a strong foundation of clear thinking based on mathematics and science.