CLOUD INNOVATION CHALLENGE · LMU × AWS
Velocity Racing × AWS
Fictional team · synthetic dataBrowser demo · no AWS calls
Overview
The fastest decision loop on the grid.
One AWS platform that turns 300+ sensors per car into race-day decisions, protects the design IP and grows the audience. Headline numbers first, then the case, then the product. Jump to the demo ↓
THE CASE IN DETAIL
What it takes to prove it.
Cost, break-even and the technical gates you would sign off. Estimates are public data and benchmarks; the pilot replaces them with Velocity’s own figures.
AWS usage €0.47 M + team €0.72 M
Priced from published AWS list prices and salary benchmarks. Break-even needs about 0.13 of one championship position. No avoided DNFs are assumed.
CFD throughput
AWS reported an F1 workflow that tripled CFD throughput and halved turnaround per run. This is evidence the approach is feasible, not a forecast for Velocity.
Read the F1 engineering story ↗Break-even calculator (illustrative)
NET VALUE (€ M) —BENEFIT / COST —BREAK-EVEN POSITIONS —
Defaults reproduce the report’s base case. Position value is a public estimate (about USD 8–12 M per adjacent position). Fractional positions are expected values (0.3 ≈ a 30% chance of one place higher). The full programme is funded only after a six-week pilot capped at €100k. No assumed DNF savings.
A pilot with technical gates
- Weeks 1–2 · Baseline and security
Landing zone, data rights and region policy. Replace the cost assumptions with Velocity’s own figures.
- Weeks 3–4 · Shadow mode
Replay a season of telemetry. Measure false alerts, missed events, forecast error and edge latency (p99) on trackside hardware.
- Weeks 5–6 · Rehearse race day
Disconnected operation, duplicate events, failover (proposed RTO 15 min, RPO 1 min) and load.
- Gate
Advance only when thresholds agreed with the CTO and race engineering pass. Then CFD and fan platforms in stages.
THE PRODUCT DEMO
See it work.
Three interactive pages on synthetic data. Everything runs in your browser; nothing is sent to a server.
01 · RACE CONTROL
Turn telemetry into a race decision
Replay a lap, trigger a thermal anomaly, change the weather and watch the strategy ranking respond. Evidence, ranges and an engineer decision record.
Open Race control →
02 · FAN EXPERIENCE
One pass, one entry
A separate public platform built from approved content. Scan the digital paddock pass twice: the second scan is blocked.
Open Fan experience →
03 · ML PREDICTIONS
Models trained and tested like the pilot
Train a component early-warning model and a tyre-degradation model in your browser, then see how they score on held-out races against the pilot’s acceptance criteria.
Open ML predictions →