ismaelloveexcel/onboarda--back-up---may-26

White-label compliance & onboarding platform — FSC Licensed, Mauritius PIS # Onboarda **AI-powered compliance onboarding for regulated financial institutions.** Onboarda automates KYC/AML due diligence for banks, EMIs, and payment providers using a deterministic multi-layer AI pipeline. It combines rule-based risk scoring, external screening APIs, and Claude AI–generated compliance memos into a single auditable workflow. The platform ships as two branded surfaces: - **Onboarda** — Client-facing portal where applicants submit documents and forms - **RegMind** — Internal back-office where compliance officers review, approve, or reject applications --- ## Tech Stack | Layer | Technology | |-------|-----------| | Backend | Python 3.11 / Tornado | | Database | PostgreSQL (production) · SQLite (local dev) | | AI | Anthropic Claude (Sonnet for LOW/MEDIUM risk, Opus for HIGH/VERY_HIGH) | | KYC Provider | Sumsub (identity verification, AML/PEP screening) | | Document Storage | AWS S3 | | PDF Generation | WeasyPrint | | Frontend | Vanilla JS — single-file HTML (no build step) | | CI/CD | GitHub Actions | | Deployment | Render.com | --- ## Repository Structure ``` onboarda/ ├── arie-backend/ # Python backend │ ├── server.py # Main API server (all endpoints) │ ├── claude_client.py # Claude AI integration (10 compliance agents) │ ├── db.py # Database layer (SQLite + PostgreSQL) │ ├── rule_engine.py # Country/sector risk scoring (FATF lists) │ ├── memo_handler.py # Compliance memo generation │ ├── validation_engine.py # 15-point memo quality auditor │ ├── supervisor_engine.py # Final review & contradiction detection │ ├── document_verification.py # Layered verification pipeline │ ├── verification_matrix.py # Document check definitions │ ├── screening.py # AML/PEP screening orchestration │ ├── sumsub_client.py # Sumsub API client │ ├── auth.py # JWT auth & rate limiting │ …