AI orchestration and RAG (Retrieval-Augmented Generation) platform for SADC organizations.
Built for hybrid dense+sparse search, multi-tenant isolation, and SADC network resilience.
# Prerequisites: Python 3.12, Podman or Docker
python scripts/verify-platform.py
make setup
cp .env.example .env # Fill in ANTHROPIC_API_KEY
make dev-up
curl http://localhost:8000/health./scripts/instant-poc.sh --prospect "Ministry of Health" --docs ./sample-docs/CLIENT_HOST=192.168.1.100 CLIENT_USER=ubuntu TENANT_ID=client-abc ./scripts/deploy-client.shThis is the consulting MVP. Full SaaS platform (billing, analytics, K8s) comes after first paid invoice.
See marketing/freelancer-packages.md for service offerings. See marketing/case-study-rstp.md for proof of capability.
- RAG Engine: Hybrid dense (BGE-large) + sparse (Splade) -> RRF fusion -> MiniLM reranking
- Multi-tenant: JWT-verified per-tenant Qdrant collections
- Resilience: CircuitBreaker (CLOSED/OPEN/HALF_OPEN) + SQLite WAL offline buffer
- Customization:
Hookable.execute()— client overrides without forking core
- Python 3.12+
- Podman 5.x or Docker
- 4GB+ RAM (for embedding model)
- Linux or macOS (Apple Silicon supported), WSL2 on Windows