I’m a Software Engineer and Data Scientist based in Munich. I build tools that bridge the gap between complex technical domains and production-ready software.

I have a PhD-shaped background in physics and materials science. I spent 2 years at the UK’s national synchrotron (Diamond Light Source) building instrumentation and deep-learning solutions for scientific imaging.

Now, I focus on building high-fidelity web applications, AI-driven workflows, and data-heavy systems.

A fully differentiable parallel-beam CT projector, reconstruction, and alignment toolkit built in JAX. Exact gradients for 5-DOF rigid-body alignment, memory-efficient streaming, and GPU-accelerated reconstruction. Probably the most technically interesting thing I've built.

• Built with: Python, JAX

A searchable, local-only clipboard history for macOS. Watches the system clipboard, archives everything into a local SQLite database, and exposes a JSON-first CLI so that agents (and humans) can recall things you've copied. Published on crates.io.

• Built with: Rust, SQLite, objc2

You've got a laptop, a pile of scrap, and zero internet. Can an LLM help you survive without getting anyone hurt? A benchmark runner with structured judge rubrics and a Next.js dashboard for exploring runs. Full writeup here.

• Built with: TypeScript, Next.js

An animated dot-matrix display component for React Native. Skia-first rendering with Reanimated for 60fps UI-thread animations, a VU meter mode for real-time audio visualisation, and a bunch of built-in presets. Inspired by the ElevenLabs UI.

• Built with: React Native, Skia, Reanimated

My most successful (and simplest) app, a notes app built around Andrej Karpathy's note-taking technique. People seem to particularly like that there's no subscription, and no AI.

• Built with: React Native Expo, TypeScript, Swift

My least successful (and most complex) app, an AI voice-notes app which is the antithesis of Gravity. People seem to particularly like not using it.

• Built with: React Native Expo, TypeScript, Railway

Before software, I spent 2 years at the UK's national synchrotron building deep-learning tools for X-ray imaging.

ML Segmentation	Deep Learning Segmentation

I'm currently taking on freelance contracts in Munich (and remote). If you have a complex project or bespoke software needs, let's talk.

• 📧 Email: tristan@tristanmanchester.com
• 💼 LinkedIn: linkedin.com/in/tristanmanchester
• 🔗 Web: tristanmanchester.com

Based in Munich, Germany 🇩🇪