Embedded Files
Phone: +1 250-580-4366
Email: dominic.p.ferreira@gmail.com
My Profile
I am a Machine Learning Engineer at Torus Biomedical Solutions Inc., where I develop and validate machine learning models for intraoperative X-ray segmentation and create synthetic X-ray images from CT data. I hold a Master’s degree in Computer Science from the University of Victoria and bring a rich background in simulation and machine learning across academic and industry settings. My expertise includes C#, Unity, Python, and advanced algorithms for motion planning and collision avoidance, with prior experience in game development.
Employment Experience
Machine Learning Engineer @ Torus Biomedical Solutions Inc.
May 2024 - Present Vancouver BC, Canada
Implemented machine learning models for segmenting intraoperative X-rays to accurately identify bone landmarks, improving surgical outcomes.
Validated model accuracy and performance using real intraoperative X-ray data, ensuring reliability in clinical settings.
Developed a pipeline for creating synthetic X-ray images from CT data, enhancing the diversity and volume of training data for models.
Created and managed pipelines for processing and labeling image datasets, essential for the effective training and testing of neural network models.
Reviewed and incorporated the latest advancements in medical imaging from current literature into project methodologies.
Utilized open-source models and datasets to develop and test machine learning prototypes efficiently.
Game Developer @ Sonical.ly
Sep 2020 - Apr 2021 Vancouver BC, Canada
Led the development of educational music mobile games, successfully deploying on both iOS and Android platforms.
Owned and designed game prototypes rapidly from the ground up, including key gameplay features: progression systems, level design, UI, UX, animation, audio, tutorials, and dialogue.
Grew the beta user base from inception to over 1000 active users, showcasing expertise in market analysis, user acquisition tactics, and responsive game development.
Diagnosed and fixed software bugs by collaborating with QA for root cause analysis, and applied timely patches for user experience enhancements.
Refactored and optimized bottlenecks found using profiling tools and resolved using best practice coding principles.
Tech Stack: C# + Unity + Git + Visual Studio + FMod + AWS
Education
Master of Science in Computer Science
2022 - 2023 University of Victoria
Honours: Graduate Award (2022, 2023) for high academic standing
Research: Graphics, AI, Design, and Games Lab Member (GAIDG)
Supervisor: Brandon Haworth
GPA: 4.0/4.0
Bachelor of Science in Computer Science and Music
2016 - 2021 University of Victoria
Publications
Deformable Elliptical Particles for Predictive Mesh-Adaptive Crowds
2024, November
In Proceedings of the 17th ACM SIGGRAPH Conference on Motion, Interaction, and Games (pp. 1-11)
Developed a more dynamic and realistic fundamental representation of simulated agents in crowd simulation
Designed a mesh-adaptive deformable agent representation method resulting in a tighter occupied space fit
Formulated probabilistic collision avoidance algorithm with accounts for rotational uncertainty of asymmetric agent geometry
Supports artists and engineers in developing diverse and accurate simulated crowds
Tech Stack: C# + Unity + Visual Studio + Windows + Python + Matplotlib + Jupyter
DeepSolfège: Recognizing Solfège Hand Signs Using Convolutional Neural Networks
2022, January
In International Symposium on Visual Computing (pp. 39-50). Springer, Cham.
Dominic Ferreira, Brandon Haworth
Developed a machine learning application to recognize musical hand signs using a webcam.
Achieved highly accurate and fast results - F1 score of 94% and real-time performance.
Created a dataset with 16,900 RGB images & domain-specific data augmentation techniques.
Published at a peer-reviewed venue (International Symposium on Visual Computing).
Tech Stack: Python + TensorFlow + Keras + Jupyter + OpenCV + CometML
Projects
SpaceCrowds
2022
Developed a ‘zero-gravity’ simulation model of humans aboard space stations
Model uses physically-based forces and torques that move agents
Designed visibility graph-based representation of navigable space
Agents have a biomechanically confined reachable workspace to interact with handles and other surfaces
Standalone application with 3D graphics, scenario editor, and visualization toolkit
Can be used for the design and safety analysis of future spacecraft plans
Funding provided by the W.E. Cowie Faculty Innovation Award
Tech Stack: C# + Unity + Visual Studio + Windows
Mixing & Mastering
Let's Chat
Google Sites
Report abuse