Daniel KillenbergerBasel, Switzerland

Daniel Killenberger · AI & software consultant · Basel

I build things.

 

Lately that means tools for working with AI agents, and a generator that grows trees from a seed. For other teams I evaluate codebases: how secure they are, how complex they have become, whether AI agents can work in them. Every finding names a file and a line. I also coach teams, in their own repo, to ship with AI coding agents. Before that I was CTO of Obligate for five years: a team, a codebase, and hundreds of millions in securities issued on it.

I used to manage people and code. Now I manage agents.

Track record
5 years CTO at Obligate, now technical advisor
Scale
Hundreds of millions in securities issued on the platform
AI agent coaching
Just days from one feature at a time to every engineer shipping several in parallel
Built and ran the tech

How I work with teams

I work with founders and engineering teams, in Basel, Zurich, or remotely, on three kinds of problem: shipping with AI coding agents through spec-driven development, code and security review, and scoping a product before it is built. Three recent cases, clients unnamed.

Spec-driven development with AI coding agents

Situation
The team was prompting one feature at a time and waiting on each response.
What I did
Across five sessions in their own repo, I introduced a spec-driven workflow and coached them through using agents on their own product.
Outcome
Within days of the first session, they were shipping in parallel.

Code and security review

Situation
The client needed to know where their software stood: how secure it was, how sound, and what to do next.
What I did
I read the whole codebase, its security and its architecture, and tied every finding to a file and a line.
Outcome
Then I wrote up what a rebuild would take.

Product scoping before the build

Situation
I was booked to coach a team on AI agents, but what the product needed to do was not yet clear.
What I did
I first helped clarify it and turned it into a spec the team could build from.
Outcome
Not what I was hired for, but what needed to be done.

Track record

Securities platformRegulated infrastructure for issuing securities on-chain. CTO for five years, now a technical advisor.obligate.comStructured products on SolanaI led the development of the initial product and its smart contracts.margarita.finance

If one of these sounds like your team, get past Jev

Things I make

On my own time, mostly because I want them to exist.

  • Forge

    Build pipeline, Grok Bot

    I mark a spec ready and walk away. Forge plans, reviews, implements and opens the pull request. It wakes me only for the decisions that are mine.

    Template
  • Oregon white oakseed 1407

    Telperion

    Procedural tree generator

    Trees grow from a seed as one structure, trunk to the twig a leaf hangs on. The same seed always gives the same tree. I am building the Nanite-like renderer that draws a forest of them in real time.

    GitHub

  • YULONG A//45

    Fast gravel frame

    An original frame, from geometry to finished model. Every number on its page says where it came from.

    Live page

  • Omarchy distraction space

    Plugin for Omarchy, a Linux desktop

    Chat and social apps stay on one workspace. Leaving a focus lock takes a written reason.

    GitHub

Get past Jev

Jev is the bot between you and my phone. Convince it, and my phone rings.

  1. Bounced
  2. Later
  3. Afternoon pile
  4. Rings my phone

Every real message reaches me. Only the good ones ring.

Or find me on . DM me or post at @DanKillenberger.