Banyan Bat: Sindhudurg — Building a Scalable API Fortress for Fraud Detection

🛡️ Sindhudurg — Building a Scalable API Fortress for Fraud Detection

When Chhatrapati Shivaji Maharaj built Sindhudurg Fort, he did it not just with stones — but with vision.
Strategically placed on the coast, Sindhudurg was a gateway, a shield, and a statement: “Swarajya watches every entry.”

So when I began building the Fraud-Free Swarajya project, the API Gateway — the place where every transaction first arrives — had to be named Sindhudurg.

🔍 The Role of Sindhudurg

This is the entry point of the system. A Spring Boot application that:

✅ Accepts payment transaction requests
✅ Calls the AI module (tanaji) over HTTP
✅ Applies fraud rules (dadoji)
✅ Stores the transaction into PostgreSQL (ramchandrapant)
✅ Caches previous scores via Redis (jiva)
✅ Exposes Prometheus metrics for monitoring (kanhoji)

All while returning a real-time fraud risk score with explanation and recommendation.

🏗️ Spring Boot Clean Architecture

We followed a layered approach:

Controller Layer — Exposes /api/v1/fraud-score
Service Layer — Orchestrates logic (AI + Rules + Cache + DB)
Client Layer — Uses WebClient to call tanaji (Python FastAPI)
DTO Layer — Separate request and response objects
Entity Layer — JPA-backed PostgreSQL persistence

🌐 Talking to AI with WebClient

Instead of embedding ML in the same codebase, we call tanaji over HTTP:


WebClient webClient = WebClient.builder().build();
Mono<FraudScoreResponse> response = webClient
    .post()
    .uri("http://tanaji:8000/score")
    .bodyValue(request)
    .retrieve()
    .bodyToMono(FraudScoreResponse.class);

Decoupling AI from business logic makes the system flexible and testable.

🧠 Redis Integration (Jiva)

To avoid re-scoring the same transaction, we cache responses in Redis:

String cached = redisTemplate.opsForValue().get(transactionId);
if (cached != null) return parse(cached);

Fast, idempotent, and aligns with real-world payment infrastructure.

📈 Monitoring with Prometheus (Kanhoji)

We expose /actuator/prometheus metrics using Micrometer. This includes:

API call latency
Cache hits
Errors

These metrics are pulled into Grafana to visualize fraud scoring patterns.

🐳 Docker-Ready Microservice

Sindhudurg runs as a container and connects with others via Docker Compose:

services:
  sindhudurg:
    build: ./sindhudurg
    ports:
      - "8080:8080"
    environment:
      - TANAJI_URL=http://tanaji:8000

✅ Sample Request

curl -X POST http://localhost:8080/api/v1/fraud-score \
  -H "Content-Type: application/json" \
  -d '{
    "transactionId": "txn001",
    "userId": "user123",
    "amount": 5000,
    "currency": "INR",
    ...
  }'

📜 Trivia & Legacy: Why the Name Sindhudurg?

Before we wrap up, here are a few fascinating facts about the real Sindhudurg Fort — which inspired the name of this module:

🏝️ Built on an island!
Sindhudurg Fort was constructed by Chhatrapati Shivaji Maharaj in 1664 on the island of Kurte in the Arabian Sea, off the Malvan coast in Maharashtra.
🧱 Over 50,000 kg of lead!
The foundation stones were secured using molten lead to make the fort impenetrable. Think of it as the 17th-century version of bulletproofing!
👣 Only fort with Shivaji Maharaj’s handprint and footprint.
This is the only किल्ला (fort) where Chhatrapati Shivaji Maharaj’s पूजास्थळ holds his actual handprint and footprint — making the fort not just strategic, but sacred.
🎯 Built for coastal surveillance.
Sindhudurg wasn’t just a fort — it was a sentinel guarding against naval invasions, much like our API guards digital transactions.
🚪 Invisible entrance by design.
The main gate is camouflaged and not visible from outside — a clever architectural move to confuse enemies, mirroring how our service quietly but smartly filters and flags risky traffic.

⚔️ Just like the original Sindhudurg stood between the kingdom and coastal invaders, this API module is the first barrier against fraud in the digital empire.

🧭 What’s Next

In the next post, we’ll dive deep into tanaji, the AI engine — including model training, scoring logic, feature extraction and a lot of fun facts about Tanaji Malusare a legend himself.

But for now — our digital fortress is operational.

🛠️ Repo: github.com/pcm1984/fraud-free-swarajya
📖 Intro Blog to start with if you are seeing this for the first time:
Building Fraud-Free Swarajya: My AI Journey Inspired by Shivaji Maharaj