Project Overview: https://vimeo.com/332002914

• Led a team of AI engineers in developing computer vision ML models for real-time train inspections, servicing hundreds of trains daily across North America’s railroad network.

• Architected and developed an AutoML pipeline, reducing model retraining cycle from multiple weeks to hours, achieving ~1M $ annual OPEX savings.

• Organized internal AI workshops to teach AI fundamentals, fostering knowledge sharing and collaboration.

• Mentored data scientists and software developers interns for 3 consecutive years.

Propose and implement AI-based solutions to improve the company’s products.

• Develop Deep Learning solutions to address business cases.

• Develop algorithms that remove noisy signals from Sonar 3D Points Cloud using Deep 3D Networks. (http://www.teledynecaris.com/en/products/caris-mira-ai/)

• The proposed solution reduces user’s processing time by more than 80%.

• Develop back-end services to serve deep learning models in the cloud.

• Research on Audio Signal Generation using Generatives Models.

• Research on Image Super Resolution using Generative Adversial Networks.

• Research on Recurrent Neural Networks for Poetry Generation. Model learns to combine French, English an Chiac.

• Research on Object Detection on real time Video Stream on edge devices.

• Design, build, train, and monitor deep convolutional networks for object detection in a real-time video stream.

• Evaluate and improve deep learning models performance (speed and accuracy) by fine-tuning models and hyperparameters.

• Apply computer vision techniques and Transfer Learning to improve models accuracy on predictions while maintaining high-speed processing.