🧠 Mackintosh Risk

Next-Generation FinTech Risk Intelligence

Built by Kayleighy Mackintosh – Mackintosh Enterprises

---

🧩 Overview

Mackintosh Risk is an advanced, auditable FinTech analytics framework that unifies credit risk scoring and fraud flagging within one reproducible R ecosystem.

It blends machine learning, real-time streaming analytics, and cloud-native data pipelines to provide CTO-level confidence in financial decision systems.

Key capabilities:

• 🚀 Credit risk modeling — XGBoost-powered PD (probability of default) scoring
• ⚡ Streaming fraud detection — Kafka-based transaction scoring with Snowflake persistence
• ☁️ Cloud storage and governance — AWS S3 Parquet event logs + Snowflake audit tables
• 🔍 Explainability — SHAP-based /explain API endpoint
• 📊 Observability — Prometheus metrics and model performance dashboards
• 🧱 Compliance-ready architecture — deterministic builds, schema validation, PII redaction

---

🧠 Architecture

flowchart LR
    subgraph DataLayer["📦 Data Layer"]
      S3[(AWS S3 Parquet Logs)]
      SF[(Snowflake - RISK Schema)]
      K[Kafka REST Proxy]
    end

    subgraph ModelLayer["🧮 Model Layer"]
      M1[XGBoost - Credit PD Model]
      M2[XGBoost - Fraud Detection Model]
    end

    subgraph AppLayer["⚙️ Application Layer"]
      API[Plumber API /health /predict /fraud /explain /metrics]
      WORKER[Fraud Streaming Worker]
    end

    K -->|Transactions| WORKER --> M2
    WORKER -->|Flags + Alerts| SF
    M1 -->|Credit Scoring| API
    API -->|Audit + Decisions| SF
    API -->|Logs| S3

Loading