Experience
Where I've worked
A full-time data engineering role, plus research and co-op roles across applied machine learning, medical imaging, and medical device software.
Machine Learning Researcher
UBC Robotics and Control Laboratory (NSERC-URSA)
- Reimplemented Stanford's EchoPrime multimodal deep-learning architecture to generate structured echocardiography reports from multi-view ultrasound videos, adapting the pipeline to UBC's proprietary clinical dataset.
- Fine-tuned a text encoder using contrastive learning on millions of echo videos and thousands of cardiac reports to improve video–report alignment, strengthening clinical interpretability and downstream retrieval performance.
Machine Learning Engineer, Co-op
- Designed a scalable, provider-agnostic LLM benchmarking framework integrating OpenAI, Google, Anthropic, Mistral, and AWS APIs to evaluate image-based IP infringement detection.
- Improved LLM image-analysis accuracy by 5–10% through prompt optimization, multi-step reasoning chains, and grounding strategies.
- Fine-tuned Gemini 1.5 Pro and Flash models on proprietary IP datasets, achieving a 5–10% improvement in intellectual property recognition accuracy over baseline models.
- Collected 80,000+ images, built a Hugging Face API pipeline, and trained ViT and hybrid ViT-CNN classifiers to detect images containing valuable assets.
- Reduced classifier latency by 75% through TorchScript compilation, knowledge distillation, and INT8 quantization, enabling real-time image analysis at scale.
Junior Software Developer, Co-op
- Developed full-stack medical device software for liver health assessment in a 3-person agile team, building a web-based diagnostic platform (HTML, CSS, JavaScript, Python, C++).
- Designed a C++/JavaScript backend with a WebSocket server for low-latency data exchange between device code and a browser-based GUI.
- Built a Python REST API to manage SQLite databases, supporting reliable CRUD operations for patient and device records.
Software Systems Engineering, Co-op
- Analyzed time-series data from cardiac ablation systems using MATLAB, developing visualization pipelines for electrical signal analysis and system performance monitoring.
- Refactored MATLAB data-handling classes, reducing self-test data-extraction runtime by 95%.
- Diagnosed and resolved reported system issues by correlating performance data with hardware and software specifications.