Services
I build and fix the practical side of AI automation: the tools, integrations, and guardrails that let agents do useful work inside your stack. Engagements start with a conversation and are scoped to what you need.
Connect agents to your repositories, APIs, and internal tools, with defined scope, approval points, and a clear record of what they did.
Purpose-built CLIs, plugins, and services that remove a repetitive step from how your team builds, reviews, or ships.
Take a promising demo, notebook, or script and turn it into something that installs, runs unattended, and can be maintained.
Diagnose agents or pipelines that loop, stall, overreach, or fail quietly, then tighten the boundaries and checks around them.
Capabilities
These are applied engineering skills, shown in public work like Dead Drop. This is not model-training research, and there are no promises that an agent will never need a human.
Supporting work
Automation is only as good as what it connects to, so the work covers the software, hardware, and hosts too.
How engagements work
There are no packages or fixed price list. Each engagement is scoped to the problem.
Message on X with the system, the goal, and the constraints.
We agree on what gets built, what stays manual, and how we will know it works.
Work happens in your repositories, in small reviewable pieces.
You get working code, the docs, and an honest list of known limits.
Start a conversation
Send a short note about the system, what it should do, and where people need to stay in the loop.