Independent AI engineering

Useful AI.
Built for real work.

Turn an idea into a product. Bring AI into your team’s daily work. Make an existing application safer. I’m Sebastian Kouba, an engineer and founder who works alongside you to ship.

Product engineeringEmbedded technical leadershipSecurity & private AI

Engineering that moves your business forward.

A focused build, a senior engineer inside your team, or a practical review of what you already run.

Find the right way to work together

Built from hands-on work.

Own products and prior engineering experience inform how I approach your project.

Preview of the fape dictation demo for Omarchy

Own product · Dictation

fape

Speak a message, note or prompt. fape puts the text into the application you are already using.

Meet fape

Product work · Coaching

Parts Room

Coaching application work exploring how AI can support reflection and help turn a conversation into a next step.

A focus on the experience, as well as the model.

Prior experience · HiQ

AI for document workflows

Private equity and invoice automation experience. A foundation for building around sensitive documents, review steps and operational constraints.

Discuss a private AI application

The person you talk to is the person who builds.

Sebastian KoubaFounder & AI engineer · Sauerlach, Germany

I’m Sebastian Kouba, founder of Tokenschmiede UG (haftungsbeschränkt) in Sauerlach. I work close to the frontier of AI tooling, with a practical question in mind: does it make the software, or the work around it, better?

I combine hands-on product development with technical leadership. You work directly with me, from understanding the problem to shipping and handing over the result.

More about Sebastian

Decisions behind the software.

A useful first version, clear data boundaries and a team that can keep moving. These choices guide the work.

Start with a useful scope

Agree on the smallest useful version and a first milestone. Test it against the actual workflow before building further.

Let data boundaries lead

Choose models, hosting and integrations around where your data may be processed. Use local or on-prem applications where the requirements call for them.

Review what gets built

AI coding agents are part of the engineering workflow. Clear tasks, careful review and verification make their output usable.

Build for the handover

Deployment, documentation and knowledge transfer belong to delivery. Your team should be able to operate the software and keep moving.

Read the engineering notes

What would you like to make work better?

Tell me what you’re building, where you’re stuck and what a useful outcome would look like. We can work out the next step together.

Let’s talk about your project

Direct contact with Sebastian. English or German.