/ ml systems
ML systems
Production inference paths with one boundary, drawn well. Reliable, observable, no surprises.
data engineer · aarhus, dk
code is a creative practice.
I build ML systems, agent workflows, and small autonomous things on a Raspberry Pi after hours — then make sharp, premium content about shipping them.
@drobeek
01
what i build
One boundary, drawn well. Production paths with no surprises — and the discipline to subtract more than I add.
/ agents
Agent workflows
Small autonomous agents that plan, call tools, and clean up after themselves.
/ edge
Raspberry Pi lab
Tiny autonomous projects running quietly on a single board at home, after hours.
writing
notes from the
after-hours bench.
Why software feels closer to art than engineering — and what I ship between posts. The technical layer of the brand, in forest green.
read the essays01 def pipeline(raw): 02 clean = normalize(raw) 03 return model.infer(clean) 04 05 # one path. no surprises.
02
selected work
Small, sharp systems from the after-hours bench — agents, simulations, and local automation built in the open.
@
PI SYSTEMS
agents · simulations · ops
Three open projects around learning agents, simulated people, and watchful automation. Small systems with visible loops and sharp edges.
see the code