Projects and Work

LLM Model for quick insight into evolving AI Regulations and laws.

At Orcawise, I joined as a Custom LLM Scientist and quickly became a leader bridging the data science and legal teams for the EU AI Act project. I played a key role in crafting project OKRs and tasks, then spearheaded the model's foundation by evaluating and selecting the most suitable base model. Leading the build process, I tackled data cleaning, training/testing splits, initial model setup, and accuracy testing. I continue to contribute significantly to ongoing refinement cycles to optimize the model's performance. My strong communication skills enabled seamless collaboration across technical and business teams, allowing me to clearly explain complex information to non-technical stakeholders. 

Application of Machine Learning to Optimise Learning Analytics

Research study carried out between 2021-2023 which studied the potential of using machine learning to support student learning at higher education institutions. My research identified most useful models for detecting patterns in student learning, identified key indicators of students at risk of non progression using non-personally identifying or sensitive information and offered actionable insights to higher education institutions regarding relevant intervention steps to boost student engagement.

Azure Machine Learning Capstone Project

In this project completed as part of a nanodegree sponsored by Microsoft in 2020, I built a supervised learning model to predict insurance claims and deployed it on Microsoft Azure. It had an accuracy of 81% at making these predictions.

Food Quality Computer Vision Model

Contributed to the development of a computer vision model using Yolo object detection toolkit to detect fresh and rotten fruits. This was a project completed as an entry to the 2021 OpenCv AI competition where we finished as finalists.

Moodle Flask API 

A research grant in 2022 was used to develop an API which supports specific data extraction from the Moodle virtual learning environment. This was shared with the Moodle community and a section was added to the Moodel documentation page via the link:

https://docs.moodle.org/403/en/Attendance_activity

Face Mask Detector Project

Trained an AI model to detect a persons wearing facemasks during the pandemic in 2020. Model was trained leveraging the Google's teachable machine platform.