AINA Data Engine And Academy Alignment Board

Current Repo Roles

Product Mission

The product mission is learner/role -> AI Fluency Capability Map -> simulation/practice -> evaluation -> proof -> next curriculum move.

From AI anxiety to AI fluency: assessment, simulation, personalized curriculum, and proof.

Title coverage, embeddings, BLS/O*NET/HF data, jobs-research assets, JD-aware role context, tool maps, evaluator rubrics, and source authority are all inputs. The product object is a learner-facing AI Fluency path that can be trusted, explained, evaluated, and improved.

Operating Goal

The active goal is reconciliation-gated: preserve the current state, fix stale metadata, finish the Fusion/core import ledger, define the export manifest, prove Academy-safe consumption, and only then resume clean data expansion.

Autonomously preserve and reconcile the AINA Data Engine Room production spine before new data expansion: create verified backup/tag proof for current head 11bf3c0, refresh the private GitHub backup if needed, fix stale board/metadata and mark embeddings parked; then complete the Fusion/core import-decision ledger, manually port only accepted diffs with gates green, consolidate local main into one branch truth, define engine_room_export_manifest_v1, and prove one platform-safe top-500/top-1,000 export can be consumed by aina-academy without live VDS/DuckDB/Python coupling, while preserving source authority, AI Fluency capability-map boundaries, exact-cosine retrieval proof, and keeping public runtime, real-user data, external writes, production telemetry, runtime embedding authority, donor mutation, and deletion blocked until explicit release receipts exist.

Current Execution Truth

Source Truth Rules

Milestones, Slices, And Tasks

The first implementation move is M0/M1: preservation proof and Fusion/core import-decision ledger. The next product move is M2/M3: define engine_room_export_manifest_v1 and prove Academy can consume a pinned top-500/top-1,000 export without live coupling.

Embedding Policy In The New Map

Gemini embeddings remain important, but they are parked until M1-M3 prove what Academy actually consumes. After that, use gemini-embedding-2 at 768 dimensions through paid Vertex ADC on project aina-495702. Exact cosine stays source-of-truth retrieval until VSS/RuVector acceleration has parity proof.

No-Write Zones And Release Boundaries

Donor repos, raw source files, real learner data, secrets, env files, credentials, billing settings, and production Cloudflare/telemetry/payment writes remain no-write zones unless explicitly scoped. Data engine builds, embedding receipts, retrieval proof, headless runtime fixtures, and Academy import dry runs remain local-only by default.

Validation Stack

cd /srv/aina/aina-data-engine-room
git status --short --branch
uv run aina-data-engine --root /srv/aina/aina-data-engine-room source-authority-start-here
uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-506-p0-gate
uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-510-retrieval-promotion-gate
uv run aina-data-engine --root /srv/aina/aina-data-engine-room production-runtime-readiness
uv run aina-data-engine --root /srv/aina/aina-data-engine-room platform-live-boundary
uv run aina-data-engine --root /srv/aina/aina-data-engine-room validate

cd /srv/repos/aina-academy
pnpm typecheck
pnpm test
bash ops/smoke/core-loop.sh

Done Means For This Mission

The next cold agent can pick up without guessing. The repo roles are recorded, milestones are recorded, older boards point to this alignment, the active goal matches the reconciliation-gated goal, and next execution starts with Fusion/core reconciliation plus the Academy export contract instead of isolated embedding or title cleanup.

Next Best Action