- CVML SDK developer- NPU Windows driver developer for Phoenix, Hawk Point & Strix

- I teach the members of the programming subteam and review the GitHub PRs to make surethat there aren’t any errors- I work with another controls mentor to improve our path following and vision tracking systems- I manage and administer the IT infrastructure consisting of 4 workstations, a server, a printer,associated user laptops, and networking gear- I also manage our SolidWorks PDM Professional deployment, including VPN access, SQL Server administration, user accounts, and cloud storage replication

- Demonstrated strong leadership skills to move from programmer to lead programmer within 1 year of joining the team- Qualified for world championships 4 years in a row, including a Carson division win in 2015- Utilized my knowledge of Java to improve our team's controls and software for the robots throughout my time as lead- Implemented the switch to the usage of the CAN bus instead of individual PWM connections and revamped our autonomous routines through a completely new path planning and following system- Improved robot performance by programming a computer vision system that isolated, tracked and automatically aligned the robot to the vision target

- Created front and back end’s for web applications to showcase various CVML projects (scene &text detection) that I helped improve- Improved scene detection & scene segmentation model robustness by adding support for various real-world image augmentations- Debugged and fixed model training, validation and testing metrics (accuracy, precision, mIOU,etc.) to correctly reflect model performance- Created scripts to automate training and testing of models, and extract models to standalone code- Implemented functionality to better visualise model inferencing results, to guide next steps inimproving model performance- Improved dataset preprocessing performance by implementing multithreading for a 15X speedup- Rewrote another dataset preprocessing script for a 43.2x speedup

- Computing and CAD TA for ENG 1P13, a 1st year engineering course at McMaster University- Supervised lab groups and assisted with the lesson being taught- Marked assignments and met with students weekly to assist them with their lab assignments- Managed my time exceptionally, between being a TA, focusing on my courses as a full-time student and other hobbies

- Migrated and updated an entire production application from GCP to AWS by myself- Updated backend from Java 7 to 11, Python 2 to 3 and Cassandra database from DSE 4.8 to 6.8- Updated all Python and Java libraries and frameworks (Tomcat, Tornado, gRPC) to their latest versions and fixed breaking changes between versions- Created Docker containers for each service to build and deploy to AWS ECS- Created scripts to migrate the entire database, and to automate the build and deploy processes.- Re-architected service-to-service communication for ECS using Service Discovery- Configured security groups for all services and the database, and an Application Load Balancer for the API- Ran e2e tests to identify any bugs or discrepancies between cloud service deployments- Achieved 4+ months of ongoing stability on AWS

- Created new front end features to add to the Skywatch platform, using TypeScript, Angular and RxJS, that also communicated with the Skywatch API- Integrated these standalone features into the development and eventually production front end in collaboration with 2 other members of the front end team- Developed a cloud & cloud shadow detection algorithm using Python and machine learning to mask out clouds from processed satellite images using only the RGBN bands- Excelled at working in a Scrum development environment

- Created web applications to showcase the various features of the Skywatch API for demonstration purposes- Implemented an image quality analyzer using OpenCV, Python and machine learning to flag irregular processed satellite images to prevent them from being sent to customers- Collaborated with 6 other team members to invoke the analyzer from the processing pipeline using AWS Lambda and Serverless