Skip to content

README.md

Latest commit

 

History

History
57 lines (45 loc) · 3.8 KB

README.md

File metadata and controls

57 lines (45 loc) · 3.8 KB

Work Groups

This directory organizes Project Tapestry work groups by durable lifecycle responsibility, not by a repeated requirements/engineering split. Each work group owns its requirements discovery, technical exploration, implementation notes, open questions, and delivery artifacts in one place.

The structure is meant to be necessary and sufficient for the current architecture: sovereign data enters the system; base models and sovereign alignment pipelines transform it; evaluation, security, and governance constrain what can ship; infrastructure and deployment make it usable.

Work group map

Work group Directory Owns
Data Governance data-governance/ Sovereign data sourcing, licensing, stewardship, residency, provenance, and contribution rights
Base Model Training base-model-training/ Adopted-base strategy, consortium training loop, shared-base continued pretraining, aggregation, and transition to consortium-owned bases
Sovereign Alignment sovereign-alignment/ Participant-owned continued pretraining, post-training alignment, instruction tuning, and portability of sovereign layers
Evaluation & Certification evaluation-certification/ Capability, cultural alignment, safety, benchmark design, certification criteria, audit evidence, and release gates
Security & Privacy security-privacy/ Privacy tiers, secure aggregation, differential privacy, TEEs, threat models, model-update leakage, and safety-preservation constraints
Infrastructure & Operations infrastructure-operations/ Heterogeneous compute, orchestration, node operations, observability, fault tolerance, platform operations, and cost/accounting plumbing
Deployment & Adoption deployment-adoption/ Serving, chat/product harnesses, integration patterns, participant rollout, developer experience, and adoption feedback loops
Governance & Participation governance-participation/ Anti-capture mechanics, contribution credit, decision rights, consortium process, and interfaces with ../governance/

Lifecycle view

flowchart LR
    DataGovernance["Data Governance"]
    BaseTraining["Base Model Training"]
    SovereignAlignment["Sovereign Alignment"]
    EvaluationCertification["Evaluation & Certification"]
    DeploymentAdoption["Deployment & Adoption"]

    SecurityPrivacy["Security & Privacy"]
    InfrastructureOperations["Infrastructure & Operations"]
    GovernanceParticipation["Governance & Participation"]

    DataGovernance --> BaseTraining
    BaseTraining --> SovereignAlignment
    SovereignAlignment --> EvaluationCertification
    EvaluationCertification --> DeploymentAdoption

    SecurityPrivacy -.-> DataGovernance
    SecurityPrivacy -.-> BaseTraining
    SecurityPrivacy -.-> SovereignAlignment
    InfrastructureOperations -.-> BaseTraining
    InfrastructureOperations -.-> DeploymentAdoption
    GovernanceParticipation -.-> DataGovernance
    GovernanceParticipation -.-> EvaluationCertification
Loading

Charter template

Each work-group README should stay short and concrete:

  • Purpose: what the group owns.
  • Why it exists: the architecture goals, ADRs, pain points, or value propositions that make the group necessary.
  • Scope: responsibilities included and explicitly out of scope.
  • Initial questions: workshop or research questions the group must resolve.
  • Early deliverables: artifacts expected before the work group can be considered operational.
  • Interfaces: other work groups it must coordinate with.