⚡ Aether — High-Frequency Financial Transaction Engine

A production-grade, event-driven financial transaction engine built for high concurrency, real-time fraud detection, and predictive load forecasting.

🏗️ Architecture

Client → FastAPI Gateway → Redis Stream → PostgreSQL Worker
                      ↓
              Analytics Engine (ARIMA Fraud Detection)

🚀 Tech Stack

Layer	Technology
API Gateway	FastAPI, Python
Auth	JWT, OAuth 2.0, RBAC
Caching & Queue	Redis, Redis Streams
Database	PostgreSQL, SQLAlchemy
Analytics	ARIMA, Z-Score Anomaly Detection
Frontend	React, Recharts, Vite
Infrastructure	Docker, Alembic

✨ Key Features

• High-Throughput API Gateway — Async FastAPI ingestion layer capable of processing 1,000+ mock transactions per second using asynchronous request handling and connection pooling
• Identity Fortress — JWT-based authentication with Redis token blacklisting, token caching for microsecond validation, and role-based access control (RBAC) to protect sensitive financial endpoints
• Event-Driven Pipeline — Every transaction is pushed into a Redis Stream immediately after ingestion, decoupling the API from the database and absorbing traffic spikes with zero lag to the user
• ACID-Compliant Worker — A background consumer pulls transactions from the Redis Stream in configurable batches and safely inserts them into PostgreSQL with full rollback support on failure
• Fraud-Sensing Engine — Time-series analysis using ARIMA forecasting and Z-score statistical detection to flag velocity spikes and unusual transaction amounts in real time
• Live React Dashboard — Full-stack frontend with login/register flow, transaction submission form, live transaction history table, fraud alert panel, and a load forecast chart

🖥️ Screenshots

Dashboard with transaction history, fraud analysis panel, and ARIMA load forecast chart

🛠️ Running Locally

Prerequisites

• Docker Desktop
• Python 3.11 (via Miniconda)
• Node.js 18+

1. Start Infrastructure

docker-compose up -d

2. Backend Setup

cd backend
conda activate aether
alembic upgrade head
uvicorn app.main:app --reload

3. Start Background Worker

cd backend
conda activate aether
python -m app.workers.transaction_worker

4. Frontend Setup

cd frontend
npm install
npm run dev

📁 Project Structure

aether/
├── backend/
│   ├── app/
│   │   ├── api/              # Route handlers (auth, transactions, analytics)
│   │   ├── core/             # Config, Database, Security, Redis client
│   │   ├── models/           # SQLAlchemy ORM models
│   │   ├── schemas/          # Pydantic request/response schemas
│   │   ├── services/         # Business logic (auth, fraud, cache, analytics)
│   │   ├── workers/          # Redis Stream consumer worker
│   │   └── main.py           # FastAPI app entry point
│   ├── migrations/           # Alembic database migrations
│   ├── requirements.txt
│   └── Dockerfile
├── frontend/
│   └── src/
│       └── App.jsx           # React dashboard
├── docker-compose.yml
└── README.md

🔐 API Endpoints

Method	Endpoint	Description	Auth
POST	/auth/register	Register a new user	❌
POST	/auth/login	Login and receive JWT	❌
POST	/auth/logout	Blacklist token in Redis	✅
POST	/transactions/	Submit a transaction	✅
GET	/transactions/	List all transactions	✅
GET	/transactions/{id}	Get single transaction	✅
GET	/transactions/stream/length	Redis queue depth	✅
GET	/analytics/forecast	ARIMA forecast + fraud flags	✅
GET	/health	Health check	❌

👤 Author

Mohit — github.com/mohitram-ck