Overview

Harberstone Commercial Bank is a mid-size commercial bank operating in the United States that provides various retail and corporate banking services. With an increase in the number of transactions, the bank was unable to perform manual AML checks due to high transaction volume; it needed a solution to reduce the number of false positives while simultaneously managing to detect the actual risk. Thus, the bank collaborated with Bacancy to develop a real-time and automated transaction monitoring system that ensures proper risk scoring and maintains an audit trail throughout.

Technologies Used

Python
Apache Kafka
PostgreSQL
Reactjs
AWS
Docker

Project Highlights

checkmark

Replaced overnight batch reviews with a real-time transaction screening pipeline built on Kafka

checkmark

Built a configurable risk-scoring engine that compliance analysts can retune without engineering support

checkmark

Added anomaly detection to catch suspicious patterns that a static rules engine would have missed

checkmark

Delivered a unified case management workspace covering investigation, annotation, and SAR filing

The Challenges

1

Manual AML reviews could not scale with growing transaction volumes: Increasing transaction volumes across branches and channels made manual, overnight reviews slow and difficult to manage, increasing the workload for compliance analysts.

2

Rules-only scoring generated too many false positives: The existing rules-based approach flagged too many low-risk transactions, creating alert fatigue and making it harder for analysts to identify genuinely suspicious activity.

3

Analysts lacked a unified workspace for alert investigation: Investigation, annotation, and case resolution were not managed in one centralized place, creating a fragmented workflow and making it harder to track alerts efficiently.

4

The existing risk model did not reflect historical risk patterns: The model overlooked factors that had historically indicated suspicious behavior, resulting in risk scores that did not always accurately reflect the actual level of risk.

Solutions by Bacancy

1

As a BFSI software development company, Bacancy replaced overnight batch cycles with a Kafka-based streaming pipeline. Transactions are screened and scored as they occur, and each event carries its scoring decision through the pipeline for complete auditability and detailed review later.

2

Our experts built the risk-scoring enginein Python on historical case data, weighting factors that had previously correlated with suspicious activity higher than routine transactions. Analysts can adjust these weights and thresholds through a configuration layer our team built, without needing a code change or engineering release.

3

To catch what static rules miss, our developers layered anomaly detection models over the rules engine. Instead of relying on fixed thresholds, it flags deviations from a customer’s typical transaction behavior, surfacing patterns that wouldn’t trip a rule but warrant review.

4

The React-based case management workspace consolidates alert investigation, analyst annotations, and case resolution into a single screen. A full activity log is tied to each case, supporting faster SAR preparation and giving compliance teams a clear audit trail.

Core Features

checkmark

Real-time transaction monitoring and scoring at point of transaction

checkmark

Configurable risk-scoring engine with compliance-controlled weights and thresholds

checkmark

Automated SAR generation with pre-assembled data for faster filing

checkmark

Full audit trail logging on every scoring decision and action

No. of Resource

04

No. of Resource

Time Frame

August 2025 - April 2026

Time Frame

Project Snapshot

Harberstone Commercial Bank

Outcomes

40% reduction in false positive alerts across the board

50% less manual review time spent per case

100% data-driven model built from re-weighted risk factors

2X faster suspicious activity report filing with full traceability

24/7 real-time screening replacing overnight batch monitoring cycles

0 engineering support needed for compliance team to retune scoring

Technical Stack

Frontend React TypeScript
Backend Python
Streaming & Messaging Apache Kafka
Database PostgreSQL
Cloud Infrastructure AWS
Containerization Docker
Security & Compliance BSA/AML aligned controlsSOC 2 aligned data handling
Project & Issue Tracking Jira

Experience With Bacancy

2500+ Projects Experienced Innovation with Bacancy!

Get access to an experienced team of developers and engineers from Bacancy, handpicked to ace your goals. Kickstart within 48 hours, no-risk trial.

Book a 30 min call

14+

Years of Business Experience

1458+

Happy Customers

12+

Countries with Happy Customers

1050+

Agile Enabled Employees

How Can We Help?