Generative AI & Full-Stack Innovation (1.5 years)

• Designed, built, and deployed one of Allstate’s first generative AI solutions aimed at enhancing the customer experience through improved communication workflows.
• The project was showcased regularly to executive leadership (VP and above), received international media attention, and was shortlisted for the 2024 Belfast Telegraph IT Awards.
• Worked as part of a cross-functional team, contributing across the full stack—leveraging Java (Spring Boot), TypeScript (React), MongoDB, Azure OpenAI, Terraform, and S3.
• Applied prompt engineering techniques, test-driven development, and agile/XP practices to ensure reliability and maintainability in a rapidly evolving space.

Natural Language Processing for Claims (1 year)

• Developed and evaluated NLP models to support claim segmentation and liability assessment at first notice of loss (FNOL).
• Built BERT-based and tree-based models using Hugging Face, spaCy, and XGBoost, applying these to structured and unstructured claims data.
• Delivered proof-of-concept pipelines and model performance analysis to support decision-making around automation feasibility.
• Partnered with domain experts and product managers to identify bottlenecks and define model success criteria.

Matplotlib’s pytest-mpl Plugin (Dec 2021 – Present)

• Maintained the pytest-mpl plugin for image comparison testing.
• Led key developments, including:
  • A new test suite and infrastructure to validate the plugin across diverse configurations.
  • HTML summary reports with interactive filtering of the image comparison test results.

OpenAstronomy GitHub Actions Workflows (Oct 2021 – Present)

• Since the inception of the project, I have played a pivotal role in its development, designing and implementing the majority of the OpenAstronomy GitHub Actions workflows for testing, building, and publishing Python packages.
• Workflows are widely used by the astrophysics community, including Astropy, SunPy, and the Space Telescope Science Institute, supporting software pipelines for the JWST.
• Maintained additional OpenAstronomy projects, including Azure Pipelines templates and build-python-dist.
• Advised projects on CI/CD best practices for testing and publishing infrastructure.

SunPy Continuous Integration Maintainer (Nov 2021 – Mar 2023)

• Managed SunPy CI/CD pipelines across Azure Pipelines, GitHub Actions, and CircleCI, evaluating and integrating new CI services.
• Configured Python testing environments and ensured comprehensive unit testing.
• Implemented new functionality, optimized existing code, and resolved bugs.
• Reviewed pull requests to uphold a high-quality, well-tested codebase.

• Secured funding and organised a Software Carpentry workshop for postgraduate researchers, fostering skills in programming and data analysis tools.
• Mentored students in Computational Physics, guiding them through Python problem-solving techniques, and ensuring effective learning for those new to programming.
• Marked and provided constructive feedback on assignments for the Mathematical Physics course.

• Worked under the supervision of Prof. Alan Hood, exploring magnetohydrodynamics equations in the context of solar physics.
• Investigated the interaction of magnetic fields with non-constant Alfvén speeds and phase mixing processes.

• Built a program to connect fish detections across video frames, improving the accuracy and reliability of the convolutional neural network model used for underwater video analysis.
• Conducted data analysis and retrained the neural network using new data, optimising model performance with better parameters.
• Developed scripts using Bash to automate tasks, enhancing efficiency in project workflows.
• Gained hands-on experience with Linux, Git, and Vim while contributing to a collaborative research project.

Shadowed the Head of Operations and colleagues in the Beams Department. In September 2013, I also served as a CERN Mini Expo Guide, where I explained particle physics concepts to primary and secondary school students in my city.