Allegra Bourne

An AI-powered wildlife monitoring project that uses a Raspberry Pi and deep learning to detect signs of respiratory illness in hedgehogs through bioacoustic analysis.
The system processes audio recordings to identify potential lungworm infections, combining IoT technology with convolutional neural networks for wildlife health monitoring and conservation. Designed as a proof of concept edge device, it demonstrates how low-cost AI systems can evolve into real-time field laboratory diagnostics with a view to introduce other modalities and extending into wider applications such as early-stage tuberculosis detection on farmland or non-invasive respiratory monitoring in formal healthcare settings.

For academic integrity reasons, this project remains private until the review process concludes. The full technical report, including architecture diagrams, dataset specifications, training methods, and performance results, will be published here once released.

Atticus is an offline, multi-model AI research companion designed to support postgraduate study, scientific exploration and everyday problem-solving. Running entirely on Windows and WSL Ubuntu with CUDA acceleration on an RTX 5080, it provides fast, private and reliable AI assistance without any cloud services.

It combines semantic intent routing, retrieval augmented generation and multimodal understanding with specialist local models including LLaVA, DeepSeek, Gemma, Qwen and Phi-3. Atticus can apply self-consistency sampling, multi-model consensus and speculative alignment to strengthen reasoning for technical, scientific and research-level queries.

The system indexes academic papers, research notes and codebases with a FAISS vector store and provides contextual answers using transparent citations and persistent memory. Designed as a personal research assistant, Atticus supports work in artificial intelligence, neuroscience, bioacoustics and wider scientific enquiry.

A comprehensive data science project analysing SpaceX launch data to predict mission outcomes and assess payload impact. Built interactive dashboards and machine learning models to determine launch success rates and provide insights for competitive analysis in the space industry.

A CV editing application built in React with a modular architecture. Supports drag-and-drop section reordering, dark/light themes with persistence, live preview rendering, and import/export to PDF, TXT, and JSON. Includes a custom parsing engine to intelligently extract structured CV data. Supports CV templates for enhanced extensibility. Grab my current CV here! (Created through laziness on my behalf and a hate of word processors.)

Interactive resume with Easter egg. Built with a hidden mystery, it demonstrates advanced web development techniques including event handling, state management, and creative user engagement. And you're looking at it... "knock several times and say who I am? (5)".