Sarah Thornton · software engineer

01

What I'm working on

featured · the brain behind this page

Neuroverse

A 3D map of everything I've built, learned and wondered about.

Neuroverse is my second brain. Claude saves what I learn, build and wonder about as small memories through an MCP server, and related ideas drift into coloured, categorised clusters you can fly through and explore. Those links and categories double as an index, so Claude can find the right memories when I ask it something.

It started with Clark and Chalmers' paper The Extended Mind, and a question: how possible is it to capture all of someone's tribal knowledge?

I like it because it sits behind all my conversations with Claude, and it's a platform for future projects powered by my own digitised memory.

skills: 3D data visualisation, hybrid search (embeddings + BM25), LLM memory extraction, MCP server design, product design
stack: React, Vite, three.js (3d-force-graph), Node + Express, SQLite + sqlite-vec, MCP
year: 2026

Code ↗

London Marathon 2027

Running 26.2 miles in April for Children with Cancer UK.

In April 2027, I'll be running the London Marathon for Children with Cancer UK, a charity that funds research into childhood cancers and does amazing work supporting young patients and their families. You can find out about their mission on their website:

Their website ↗ My fundraiser ↗

02

About

I'm a software engineer working on AI and machine learning. Lately I'm most interested in where AI meets neuroscience, philosophy and linguistics: how minds, memory and language actually work. I love making my projects plug into each other, which is why this site borrows its background from one of them. Away from the screen, I'm usually reading or with friends.

03

How I got here

2016 My first sale I sold custom HTML/CSS pages for a premium currency of a game which I loved at the time. They were pretty shoddy - but did the job!

2021 — 2025 BSc Computer Science · Newcastle University My dissertation was developed in collaboration with VinDo Technology: “Developing Offline AI Models for Large Scale Confidential Data Processing”. I compared ways to run NLP offline on CPU-only machines over 1TB of data, experimenting with pruning, quantisation, knowledge distillation and small transformers built from scratch. I graduated with a first in my degree and enjoyed my final university summer abroad after a looong year of applying to grad schemes.

2023 — 2024 Year in industry · Rolls-Royce Placement year across data platforms and cyber architecture. Designed a deployment workflow that saves about £50k a year, and was the first intern company-wide to receive a performance bonus. I loved my time at RR and made friends for life who I still speak to today.

2025 — now Software Engineer, AI/ML · LSEG Machine learning and applied AI: explainability for a risk-screening matching model (XGBoost, SHAP), and an AI system that writes architecture docs and answers questions about our systems in plain English. I also reverse mentor members of the Executive Committee on AI topics.

04

Say hi