Rylen Sampson Email & Phone Number

Toronto, On, Ca

Kingston, On, Ca

Kingston, Ontario, Canada

• Conduct statistical analysis for projects to assess significance and impact.• Leverage natural language processing techniques to support qualitative and thematic analysis on a variety of projects.• Support the maintenance of a client dashboard with Tableau.• Visualization of data for use by consultants and clients using Tableau, R, and Python.

Burnaby, British Columbia, Canada

• Researching how data augmentation can solve the data sparsity challenge in data-to-text generation.• Adapted and tested various data augmentation strategies from other tasks for data-to-text generation.• Through data augmentation, improved upon two state-of-the-art models whilst using less training data.• Approved thesis on ‘Data Augmentation for Text Generation from Structured Data’.• Provided data analysis support for an ongoing electric vehicle charging project on campus.

Toronto, Ontario, Canada

• Improved internal processes to support consultants by fine-tuning a large language model (LLM) with climate finance text data.• Helped deliver weekly insights to the company based on client data.• Accepted paper at “NeurIPS 2022, Tackling Climate Change with Machine Learning” workshop.

Kingston, Ontario, Canada

• Developed a deep learning model for predicting the informativeness of tweets based on their writing style to natural disaster relief organizations.• Developed experimental code, testing various iterations of the recurrent neural network model.• Achieved an improvement on results comparative to research with similar experiment set-ups. The proposed model improved on metrics such as accuracy and F1-score.• Second author on a paper from the resulting work published in “Elsevier: Information Processing & Management”• Worked under the supervision of Dr. Ding.

Kingston, Ontario, Canada

• Explored the application of machine learning methods - namely Gaussian Processes (GPs) - for constructing a potential energy surface (PES) of a molecule.• The implemented GPs addressed problems such as high dimensionality and overfitting. Also reduced computation time which is a main barrier to PES calculation.• Summarized methods and findings in an end-of-term report for future lab use. The results largely focused on the PES for hydrocarbons.• Worked under the supervision of Dr. Carrington.

Kingston, Canada

Dartmouth, Nova Scotia

Bedford, Nova Scotia

Master'S Degree, Computer Science

Activities and Societies: SFU Sports Analytics Seminar Series; Resilient Grids Team Data-to-text Generation Data Augmentation Human.

Bachelor Of Science - Bs, Mathematics And Statistics

Activities and Societies: Queen's Mostly Autonomous Sailing Team; Queen's Satellite Engineering Team; Co-President of the Mathematics &.

High School Diploma, Sciences

Activities and Societies: Varsity soccer, Soccer coach, Math tutor