Designed and developed a full-stack web application from scratch for data collection targeting software developers in North America and Europe, ensuring adherence to PIPEDA and CPPA regulations

• Built a React/Redux based Single Page Application (SPA) with protected routes (JWT authentication)
• Deployed a Node.js/Express.js/MySQL API server using Docker and PM2, ensuring 100% server uptime
• Achieved 10x increase in user traffic and collected data by integrating SurveyMonkey and Prolific APIs
• Screened and recruited 56 professional open-source developers and managed their compensation
• Technologies: MySQL, Docker, Node.js, PM2, React.js, Python, Express.js, TypeORM, Axios, JWT

Led a mixed-method study on human-bot interactions on GitHub, uncovering new insights into perceptions of bot autonomy and personality. Published one ICSE paper and successfully defended thesis

• Used machine learning and statistical methods for quantitative data analysis
• Collaborated with six researchers across three universities in Canada, Netherlands, and Denmark on a weekly basis
• Twice awarded as the "Top-Performing Graduate Student", receiving a total of 2,500 CAD in prizes

• Instructed in lab sessions
• Compiled teaching materials and conducted regular office hours
• Evaluated assignments, supervised exams, and implemented Python-based auto-graders for the final project

• Graded assignments, exams, and final course projects
• Contributed to improving the Department of Reuse format checker

Developed supply-chain solutions using Genetic Algorithm for logistics and scheduling of cargo ships in real-world scenarios with dynamic constraints. Optimized the application to run 60x faster. Wrote more than 80 unit tests

• Employed data-driven, cache-aware, and multi-threaded design
• Wrote an AMPL language parser for testing solutions and benchmarking
• Built a real-time telemetry system to monitor the application's runtime via a webpage
• Technologies: C++17, CMake, GTest, GDB, STL, Perf, SSH, Ubuntu, Node.js, p5.js, Socket.IO, HTML, AMPL

Developed Computer Vision pipelines to classify local rice varieties with 87.1% accuracy. Trained Deep CNNs to detect broken or corrupt rice seeds with 97.6% accuracy. Reduced classification time by 120x compared to human baseline

• Used OpenCV with the Canny algorithm to automate seed isolation and cropping
• Curated a labeled dataset of 1M+ rice seed images with variety and quality metadata
• Utilized TensorFlow to train, test, and deploy Deep CNNs + Softmax classifiers
• Technologies: Python, OpenCV, TensorFlow, Keras, Jupyter Notebooks, Ubuntu