• My name is Rob and I am an experienced Data Scientist with 4 years of experience in the MedTech industry, working as a Senior Data Analyst creating and wrangling large machine datasets to reduce costs, carbon emissions, energy, and water consumption. Prior to this I worked 4 years in the pharmacutical industry as a Project Engineer.
• In my current role as a Machine Learning Engineer at PitchedIt, my work involves using Computer Vision, Deep Learning, NLP, traditional Machine Learning and collaborating with other software engineers and business leads to reduce poor applications on the platform.
• I describe myself as a self starter and team player seeking to help drive innovative solutions with Data Science and ML.
• With a first-class honours Data Science Master’s degree and published research, I can address intricate AI challenges and provide effective solutions.

Whether you need a self-starter for AI readiness or a team player to enhance and maintain systems, my examples illustrate my abilities to tackle complex AI challenges and offer effective solutions for data-driven decision-making, intelligent applications, and valuable insights.

• I used Deep SHAP to explain a sub-component of deep Q learning called Experience Replay and show how its size can be tuned to improve efficiency of a Reinforcement Learning algorithm.
• Was honoured as the first in TU Dublin's School of computer science to publish in a journal during an MSc.
• Published this research and presented it at a conference in Lisbon, leading to it being featured on the cover of Machine Learning and Knowledge Extraction scientific journal contributing to the ML community.
• Thesis, Paper, Paper, GitHub
• keywords: Deep Q-Learning, SHAP, deep learning, vision, pyTorch, model optimisation

• Use YOLO a pretrained deep learning computer vision model to detect players and ball in video clips. Go frame by frame to track their changes. Extract stats such as % ball possession, player formations, etc to create a 2d pitch allowing user to playback key events. Pass this as context for ChatGPT via API to allow user to query.
• GitHub
• Keywords: YOLO, deep learning, vision, llm, django

• Built a flask application that integrates with ChatGPT, DALLE and Microsoft 365 ecosystem.
• Demonstrated it and underlying theory of multimodal large language models to US based lawyers
• GitHub, Video Presentation
• keywords: mm-llm, python-pptx, Tkinter

• Web crawled the Irish Revenue Commissioner’s website and extracting texts to create a training corpus
• Built a flask application to allow corpus to be accessed by ChatGPT.
• Presented it and underlying theory to members of the Computer Division of Engineer’s Ireland
• GitHub, GitHub, Video Presentation
• keywords: llm, beautifulSoup, Flask, threading, xml, API

• Explored data and built a collaborate filtering-based recommendation system by feature engineering marketing metrics and training a different ML models to predict what products customers will buy in the next 7 days.
• GitHub
• keywords: seaborn, scikit-learn, random forest, svm, knn, hdf5

• Built a virtual training environment using Tkinter where user can draw sand to represent roads
• Trained a Deep Q-Learning agent to navigate from top left to bottom right without driving on sand
• Presented it and underlying theory to a Dublin based Computer Club called Tog
• GitHub, Video Presentation
• keywords: numpy, pyTorch, matplotlib, Kivy, deep learning, vision

• Received a first-class Honors (1.0) in both.
• Published research in reinforcement learning and presented it at a conference in Lisbon, leading to it being featured on the cover of a scientific journal and helping contribute to the ML community.

• Received a 2.2 in both, accepted into Trinity College Dublin’s research dept. focusing on tribology, 3d printing and laser cutting.
• Top 15% finish at Silverstone racetrack UK with a student-built formula 1 race car. My work was on data acquisition and telemetry from the cars sensors to inform the team on performance and possible track strategies.