T Matsika

Teaching Assistant at Queen's University (AI4Good Lab)

Mentored a group of students through the full AI Capstone lifecycle, including supervised/unsupervised learning, deep learning, and sequence models. Provided technical guidance on ML model selection, data preprocessing, and deployment.

Applied AI Program Design & Delivery Intern at Vector Institute

Supported the development of applied AI training programs for industry and academic audiences. Analyzed training outcomes and produced reports evaluating program effectiveness. Coordinated multi-week training initiatives and collaborative projects.

Teaching Assistant at Queen's University

Supported lab-based instruction, technical troubleshooting, and grading for multiple undergraduate Electrical & Computer Engineering courses, including embedded computing and programming.

Graduate Research Assistant at Queen's University

Conducted master's research on applying machine learning to optimize network traffic in HPC environments using SmartNICs. Designed and evaluated lightweight predictive models for real-time traffic prediction on constrained hardware. Contributed to two IEEE publications and presented findings at international conferences.

AI Program Design & Delivery Intern at Vector Institute

Assisted in the development and management of AI upskilling and education programs, including talks series, multi-day workshops, and multi-week training; designed and delivered hands-on sessions and mentoring; provided guidance on ML concepts.

Teaching Assistant at Queen's University

Supported lab-based instruction, technical troubleshooting, and grading for multiple Electrical & Computer Engineering courses. Helped students with foundational topics including systems design, embedded computing, and programming.

Graduate Research Assistant at Queen's University

Designed and deployed machine learning models for real-time network traffic prediction in HPC environments. Built an end-to-end ML pipeline (data collection, feature engineering, training, evaluation, deployment). Optimized Random Forest models for deployment on NVIDIA BlueField SmartNICs under tight memory and latency constraints. Achieved ~90% precision and recall while maintaining real-time inference performance. Reduced inference latency and memory footprint for on-device execution. Published results in IEEE Cluster Workshops and EuroMicro PDP.

Machine Learning Fellow at AI4 Good Lab

Led sessions reinforcing foundational ML concepts, including supervised/unsupervised learning, deep learning, and sequence models; provided technical guidance on ML model selection, data preprocessing, and project development; mentored students to build confidence and deliver outcomes.

Machine Learning Fellow at AI4Good Lab

Built CNN models to extract education-related indicators from satellite imagery. Developed geospatial visualizations to highlight regional disparities in education access. Translated remote sensing data into actionable insights for policy stakeholders.