Image Classification on a Raspberry Pi 5 With ONNX Runtime

Evaluation suites grow faster than the codebase they cover. We treat them as first-class artefacts: versioned, reviewed, and regenerated on a schedule. The team that owns the model owns the eval set, not a separate QA group.

Open questions

When the system is wrong, the user should be able to understand why in under thirty seconds. Citation links, confidence scores, and the exact retrieved passages are surfaced in the UI for every generated answer.

Retrieval quality is the lever that moves the most weight. No amount of prompt engineering compensates for a retriever that consistently surfaces the wrong passages. We spent two weeks tuning chunking and reranking before touching the prompt template.

What we changed

Hardware is a moving target. The Jetson Orin we benchmarked in January was outperformed by an off-the-shelf mini-PC by August. We re-run the benchmark matrix every quarter and have stopped making long-term hardware commitments.

Documentation written by the team that builds the system tends to be more useful than documentation written by anyone else. The trade-off is consistency, which we address with a shared style guide and a lightweight review process.

Cost modelling is now part of our pre-merge checklist. Every PR that touches an LLM call includes an estimate of the per-request token spend and the expected daily volume. Surprises in the monthly invoice have dropped to nearly zero.