Daryl Fung

Reduced LLM training cost by over 90% and built end-to-end ML pipelines. Architected an MCP-based AI platform and developed agentic AI automation reducing manual verification by 89%.

Developed transformer and deep learning models for turbulent flow forecasting and super-resolution, improving RMSE loss by over 10%.

Harmonized 500+ health datasets across 10+ institutions for Health Data Research Network Canada.

Led machine learning tooling for AI music generation from prototype to production, improving inference speed by 43% through model optimization.

Built Wav2Lip talking-head generation, a heart ventricle segmentation model contributing to FDA approval, and AWS-hosted model serving with load balancing.

Deployed NLP and speech-processing ML toolkit as Docker microservices, doubling investor demo accessibility and improving onboarding speed.

Improved deep learning algorithms for longitudinal gut microbiome analysis, achieving 0.17 AUC improvement in disease severity classification.

Developed deep reinforcement learning motion planning and a custom autonomous driving simulation environment spanning 10+ scenarios.

Developed a DialogFlow NLP chatbot and automated email reporting system.

Developed deep learning voicemail NLP and Word2Vec / TF-IDF clustering models.

Developed GPU-accelerated NLP models.

Developed deep reinforcement learning optimization and motion plus computer vision systems for FIRA RoboCup 2017.

Master of Science in Data Science with a specialization in Artificial Intelligence in Bioinformatics, completed from 2021 to 2022.

Bachelor of Science with Honours in Computer Science, specializing in Artificial Intelligence, completed from 2015 to 2020.