Daniel MangelsdorfMelbourne, Australia
Illustration of an astronaut’s gold visor reflecting the lunar module and the Moon’s surface

Apollo 11

Interactive Lunar Landing Experience

Explore experience About the project →

Interactive demo

Research projects

Explore all projects ↗

Image generation / Model experiment

Image 1

Existing recording of candidate selection and image expansion. This prototype plays recorded output only.Recorded walkthrough

Extend an image beyond its borders and compare candidate completions from a diffusion model.

Language models / Training experiment

LLM Maker

Existing replay of a recorded generation log. It is not a live conversation with the model.Recorded walkthrough

Train small language models on one GPU, with resumable runs and an inspectable training history.

Generative geography / Model experiment

Map 1

Existing flyover recording of map tiles rendered as a suburban scene. Rendering is separate from generating the map program.Recorded walkthrough

Represent geography as a program and explore generated map tiles in a 3D renderer.

Further experiments

Diffusion Explained ↗Site Safety Detection ↗
The complete project index →

About

Hands-on building.
Clear explanations.

I’m Daniel, based in Melbourne. My work spans browser applications, data tools and experiments with machine learning.

My professional background is in technical pre-sales at Dassault Systèmes: discovery workshops, proof-of-concept environments and explaining how systems fit into existing workflows. The projects here focus on work I can show publicly.

I’m interested in training small models on targeted datasets and understanding what they can—and cannot—do.

Contact

Reach out if you're interested in AI