AI engineering · Real technical proof
Can an AI know when not to act?
An inspectable safety check for agentic systems, with clear human authority boundaries.
Try the demonstrationAI engineer · Creative builder · Berlin
I build useful things.
And little worlds.
I turn messy problems into thoughtful, testable products. Sometimes they’re AI systems. Sometimes they’re stories, experiences or everyday objects. The thread is care.
01 / Selected workThree doors, not thirty tabs
A few things,
made real.
A good project should be easier to try than to explain. These three show the different sides of how I think and build.
PURNAM · EVERYDAY RITUALS 02 / BRAND
Brand building · Objects & everyday essentials
What if everyday rituals belonged to one world?
One PURNAM brand linking STILL Objects’ tea Ritual Studio with CREA+ — a research-led daily essentials concept.
Enter the PURNAM world
HYPERSPACE · OUT OF PANEL 03 / PLAY
Interactive storytelling · Playable
What if a story could be held?
A small world inside a coin. Turn it, look closer, and help return it to its story.
Step inside the storyThere’s more in the studio. I keep the technical evidence and other experiments available without putting every detail in your way.
All AI engineering proofs
02 / How I work
Make it useful.
Make it useful.
Then prove it.
I start with a real problem, build something people can use, and check whether it actually helps. With AI, that means tests, clear boundaries and a person who remains in charge.
UnderstandMakeVerify
Want the engineering details?
I work with Python, FastAPI, RAG, MCP, and React/TypeScript. My deeper case studies show evaluations, tests, limitations and release decisions instead of relying on claims.
Read the technical work ↗03 / Beyond the CV
There’s a person
behind the projects.
Food, art, martial arts, meditation, unfinished stories and a fair bit of curiosity. The work is part of my life, not all of it.
Wander into the garden04 / A conversation