Roadmap
16-week v1: credibility baseline first.
Schedule is planned and will update. Principle: virus minimal loop → three scenarios → Unity + strategy search → Unreal + regression hardening.
§ 1
Phase plan
- Phase 0 · 2 wksPlanned
Virus minimal loop
Core + replay + calib/UQ/L1/V-lite + minimal Web player
- Phase 1 · 4 wksPlanned
Three scenarios live
Queues, ablation, audit trail, digitization-error toolchain
- Phase 2 · 4 wksPlanned
Unity + AI strategy
Report-linked replay; virus strategy search
- Phase 3 · 6 wksPlanned
Unreal + regression
Performance, golden cases, report templates
§ 2
v1.5 / v2 / v3
Deepen five AI paths: knowledge graph, multimodal, structural simulation, human data, DMTA — none are v1 acceptance promises.
§ 3
Risks & mitigations (summary)
| Risk | Mitigation |
|---|---|
| Digitization error contaminates calibration | Error bands + multi-source cross-check |
| Wide intervals gaming coverage | Mandatory sharpness |
| Scope creep (molecules/hardware) | v1 hard boundaries + ADRs |
| Replay size blow-ups | Performance budget + shard protocol |
§ 4
Team-size assumption
Assumes a small cross-functional group (simulation / backend / frontend-or-client / scientific writing). Headcount TBD — this site does not invent people or CVs.