Mobile architecture in AI era
MVVM was the right answer while humans typed the boilerplate. Agents type it now - so my solo apps run the strict architectures built for hundred-engineer teams. Receipts from Era and Lekton.
Notes from the workbench
What we're building,
and what it taught us.
Working notes from Erasmus Labs - engineering, agents, and the products I'm shipping.
-
-
What a DGX Spark is actually for
Post 1 said rent the GPUs. The asterisk: renting is for inference. For finetuning, a $3k box with 128GB beats the math - and the meter changes what you try.
-
Buy the server, rent the GPUs
Clouds overcharge for the boring server, retail overcharges for the GPUs. So I self-host Era's backend on a hand-me-down Xeon and rent the inference. The actual 2026 numbers, plus the recommendations to get started building.
