May 2025 – Present · 1 yr · Toronto, Ontario, Canada · On-site

• Engineering production-scale generative AI hair virtual try-on system using Stable Diffusion models with LoRa adapters, serving beauty recommendation features to millions of users across L'Oréal's digital ecosystem
• Architected distributed ML inference pipelines with microservices architecture, reducing model serving latency by 40% through optimized preprocessing, batching, and caching strategies
• Developed critical data preprocessing pipeline including image resizing, face swapping, background/foreground segmentation, and hair mask generation, contributing 2,500+ lines of peer-reviewed code (10% of codebase)
• Built and deployed containerized ML services using Docker and GCP with automated CI/CD pipelines for seamless production deployment
• Collaborated with cross-functional teams of product managers and designers to integrate AI features into consumer applications, following Agile development practices
• Implemented advanced computer vision techniques including inpainting for seamless hair style integration and real-time image processing

Jun 2024 – Sep 2024 · 3 mos · Remote

• Architected comprehensive end-to-end ML pipeline processing 200,000+ medical images, achieving 1.5x improvement in model precision and recall through advanced data preprocessing and feature engineering
• Developed sophisticated image similarity algorithms utilizing cosine similarity and deep feature extraction for intelligent medical data curation, enhancing dataset quality and model training efficiency
• Built production-ready ResNet-based classification models achieving exceptional 98%+ precision and recall across complex 3-way medical image classification tasks
• Implemented ML solutions using PyTorch, PyTorch Lightning, and Scikit-Learn with data analysis using Pandas and NumPy for scalable medical imaging applications
• Applied advanced computer vision and deep learning techniques to solve real-world healthcare challenges in medical image analysis