feat: bootstrap HoopScout scouting app

docs/architecture.md

@@ -1,13 +1,30 @@
 # Architecture
 
-Describe the project architecture here.
+HoopScout is a private scouting application composed of three local Docker services:
 
-Include:
+- `frontend`: Angular single-page application for player search and profile review.
+- `backend`: Django REST Framework API with Django admin for restricted data management.
+- `db`: PostgreSQL database for users, leagues, teams, players, season stats, and game logs.
 
-- main components;
-- runtime dependencies;
-- data flow;
-- persistence;
-- external integrations;
-- deployment topology;
-- relevant ADRs.
+## Data Flow
+
+Users authenticate through Django session/basic authentication. The Angular application calls `/api/players/` with search and filter query parameters. The backend returns paginated player summaries ranked by efficiency and points, plus detailed player profiles at `/api/players/{id}/`.
+
+## Persistence
+
+The initial schema stores:
+
+- users and user profiles with `admin`, `scout`, and `viewer` roles;
+- male player identity, bio, position, optional role, measurements in cm/kg, nationality, current team, and external source metadata;
+- leagues and teams, with an initial focus on European leagues;
+- one active season;
+- per-season averages, totals, and advanced metrics;
+- per-game logs for best and worst performance views.
+
+## External Integrations
+
+No automated external ingestion is included in the MVP. Demo data is synthetic. RealGM, Proballers, or similar data providers require a later authorized API/import decision before real data is collected.
+
+## Relevant ADRs
+
+- `docs/adr/0001-bootstrap-stack-and-data-boundaries.md`