// architecture | eval readiness
From €300
One focused session on architecture, eval readiness, or serving risk — written notes included.
- Live call or async video
- Baseline metric snapshot
- Prioritized next steps
- model fit note
// fixed rates
// pricing · measuredModel work with acceptance metrics
No “AI strategy” retainers without evals. Every sprint names the metric that must move — val loss, golden accuracy, or serve p95.
// two weeks · one measured outcome
€7,500
Eval harness, fine-tune program, or serve path — acceptance metrics locked before day one.
- Scoped deliverable + gates
- Loss / eval tracked
- Private GPU path if needed
- Handoff your team can own
// model ops continuity
€6,000/mo
Ongoing partnership for regression gates, retrains, and router tuning after the first ship.
- ~8 days / month
- Direct lab line
- Eval suite ownership
- 3-month minimum
// payments per contract · GPU cloud hours / hardware billed separately when out of scope