Financial Services

Real-Time Fraud Detection Architecture

Mission Briefing

Score transactions in milliseconds: card events → Kinesis → Lambda feature enrichment → SageMaker real-time endpoint → decision in DynamoDB, with async case management via SQS. Mirrors the BleepxLab fraud project — now as production cloud infra.

How it fits together

Follow the request/data flow through the architecture.

Kinesis

→

AWS Lambda

→

Sagemaker

→

Amazon DynamoDB

Concept Walkthrough

0/4 understood

Work through each concept and mark it understood. This unlocks the knowledge check.

Kinesis

Kinesis is a key service/concept for this mission.

Why it matters

Understanding it is part of architecting this solution end to end.

When to use it

Research where Kinesis fits in the architecture above and how it connects to the other services.

Hands-on: Infrastructure as Code

A realistic main.tf for this mission. Read it top to bottom — it mirrors the concepts above.

hcl

# Real-Time Fraud Detection Architecture
# BleepxCloud — FINANCE · Advanced
# Realistic starter — adapt names, regions and CIDRs to your environment.

terraform {
  required_providers { aws = { source = "hashicorp/aws", version = "~> 5.0" } }
}

provider "aws" {
  region = "eu-west-1"   # pick a region close to users / data-residency rules
}

# FinOps: mandatory cost-allocation tags on every resource.
locals {
  common_tags = {
    Project    = "bleepx-cloud"
    Domain     = "finance"
    Mission    = "realtime-fraud-arch"
    ManagedBy  = "terraform"
    CostCenter = "REPLACE_ME"
  }
}

data "aws_caller_identity" "current" {}

# Event-driven function fronted by API Gateway.
resource "aws_lambda_function" "api" {
  function_name = "bleepx-api"
  runtime       = "python3.12"
  handler       = "app.handler"
  filename      = "build/api.zip"
  memory_size   = 256          # CPU scales with memory
  timeout       = 10           # hard ceiling is 900s (15 min)
  role          = aws_iam_role.lambda.arn
  environment { variables = { TABLE = aws_dynamodb_table.app.name } }
}

# Single-table design with on-demand billing and a GSI.
resource "aws_dynamodb_table" "app" {
  name         = "bleepx-app"
  billing_mode = "PAY_PER_REQUEST"   # no capacity planning
  hash_key     = "pk"
  range_key    = "sk"

  attribute { name = "pk"; type = "S" }
  attribute { name = "sk"; type = "S" }
  attribute { name = "gsi1pk"; type = "S" }

  global_secondary_index {
    name            = "gsi1"
    hash_key        = "gsi1pk"
    projection_type = "ALL"
  }
  point_in_time_recovery { enabled = true }
}

Knowledge Check

Finish the walkthrough above to unlock the check.

Cross-verse learning

Bleepx skills build on each other. Continue your learning path across Query, Lab, and Cloud.

BleepxQuerypractice

Query the data that powers cloud services in BleepxQuery.

BleepxLabnext

Explore datasets in BleepxLab before moving them to the cloud.

← Previous All FinOps missions Next →

Hands-on: Infrastructure as Code

A realistic main.tf for this mission. Read it top to bottom — it mirrors the concepts above.

hcl

# Real-Time Fraud Detection Architecture
# BleepxCloud — FINANCE · Advanced
# Realistic starter — adapt names, regions and CIDRs to your environment.

terraform {
  required_providers { aws = { source = "hashicorp/aws", version = "~> 5.0" } }
}

provider "aws" {
  region = "eu-west-1"   # pick a region close to users / data-residency rules
}

# FinOps: mandatory cost-allocation tags on every resource.
locals {
  common_tags = {
    Project    = "bleepx-cloud"
    Domain     = "finance"
    Mission    = "realtime-fraud-arch"
    ManagedBy  = "terraform"
    CostCenter = "REPLACE_ME"
  }
}

data "aws_caller_identity" "current" {}

# Event-driven function fronted by API Gateway.
resource "aws_lambda_function" "api" {
  function_name = "bleepx-api"
  runtime       = "python3.12"
  handler       = "app.handler"
  filename      = "build/api.zip"
  memory_size   = 256          # CPU scales with memory
  timeout       = 10           # hard ceiling is 900s (15 min)
  role          = aws_iam_role.lambda.arn
  environment { variables = { TABLE = aws_dynamodb_table.app.name } }
}

# Single-table design with on-demand billing and a GSI.
resource "aws_dynamodb_table" "app" {
  name         = "bleepx-app"
  billing_mode = "PAY_PER_REQUEST"   # no capacity planning
  hash_key     = "pk"
  range_key    = "sk"

  attribute { name = "pk"; type = "S" }
  attribute { name = "sk"; type = "S" }
  attribute { name = "gsi1pk"; type = "S" }

  global_secondary_index {
    name            = "gsi1"
    hash_key        = "gsi1pk"
    projection_type = "ALL"
  }
  point_in_time_recovery { enabled = true }
}