- Worked closely with other committee members to boost attendance at events. Anticipated their needs and took initiative, designing and distributing posters and other media prior to events.
- Adapted my schedule when old projects changed and when new projects started.
- Built responsive website with member and events management system using HTML, CSS, PHP, SQL and iCalendar as well as Slack and Mailchimp integrations to allow the committee to manage the society better. Used Google Analytics to monitor traffic and assess effectiveness of website content. Made adjustments accordingly.
- Co-founded and produced a science podcast, managing the technical aspects including the website, radio station, emails, RSS feed and editing and mastering the raw audio. Analysed email and website click-through rates to boost listening figures. I personally interviewed guests on occasions.
- Assisted with funding applications by recommending new podcasting equipment as well as participating in interviews for society awards. The society won 2nd most innovative society in St Andrews in 2017 primarily for the podcast.
- Implemented new advertising strategies including advertising through a Snapchat Geofilter and digital displays. Improved the society’s social media presence by migrating to a Facebook Page.
Experience
ALLSTATE NORTHERN IRELAND
Data Scientist
Sep 2022 — Present
UNIVERSITY OF ST ANDREWS PHYSICS SOCIETY
Treasurer | Publicity Officer
May 2016 — Apr 2019
HUMBOLDT UNIVERSITY OF BERLIN
Student Intern
Jun 2017 — Aug 2017
- Spent three months during Summer 2017 working on a project involving using a convolutional neural network to detect, from videos, fish swimming on the surface of sulfur water.
- Wrote a program that takes the fish detected by the neural network and connects the fish across frames in the video. It then filters out poor quality detections by setting a minimum number of frames a chain of detections has to be present in before it is considered to be a fish.
- My program significantly improved the reliability of the output from the neural network. This was proven by comparing the filtered detections with the detections from the neural network using annotated ground truth frames.
- Studied neural networks and retrained the existing neural network with new training data and different training parameters. I created multiple models then analysed their accuracy to find the optimal parameters.
- Used the Linux command line extensively and created Bash scripts to automate many tasks. Developed my skills using
git
andvim
.
STUDENT AT ST COLUMB'S COLLEGE
Peer Mentor and Prefect
Jan 2014 — Jun 2015