Guzii · agentic AI engineering
I build tool-connected agents, multi-agent workflows, and practical automation, along with the software, radio, and Linux infrastructure they run on.
What I build
The interesting part is the edge where a model meets a repository, an API, a radio, or a person who has to approve something.
Agents that read repositories, call APIs, and act through real tools, with scope and approval boundaries set before they start.
Persistent coordination between agents across machines, providers, and people, including handoffs, reviews, and escalation built in.
Python and TypeScript services, Linux hosts, LoRa radios, and edge-to-cloud pipelines. The automation has to land somewhere.
Selected work
A private mailbox where autonomous coding agents exchange tasks, reviews, and reusable skills across human teams, machines, and providers, so work keeps moving without anyone relaying messages.
Hermes plugin support is experimental.
Open-source software that turns retired LoRa gateway hardware into a browser-managed Meshtastic and MeshCore base station that hears everything in range and can transmit too.
Smart relay is experimental.
The cloud layer that combines what many Meshpoints hear into one searchable, network-wide view, with live maps, topology, packet analytics, and fleet health.
How the work runs
Agents are useful when you can tell what they're allowed to do, what they did, and when they'll stop.
Start a conversation
Send a short note about the system, what it should do, and where people need to stay in the loop.