endorsedCompleted

Hadoop-based Fraud Detection for Banking Institution

Services Covered:

Big Data Analytics and Business Intelligence • AI and Machine Learning

Industry

Financial Services

Duration

45 months

Budget

10K - 49K

Start Date

1 June 2018

Team Size

1-5 Employees

Engagement Model

Onshore

Project Capability Score (PCS)

75/100

Good

Project Capability Score (PCS) estimates how capable Profinit is of successfully delivering a project like this, based on its past experience, track record, and ability to handle similar work. Learn more about PCS

Client

Ceska sporitelna

Czech Republic

Project Summary

Profinit provided consulting and Hadoop development services to a banking institution (Ceska sporitelna) to integrate payment transaction sources and build analytical models for fraud detection. They implemented data integration into a Hadoop-based data mart, built analytical models using Spark and Python, and integrated the models into an online transaction monitoring front-end. The engagement led to a significant decrease in fraudulent transactions, improved client NPS, effective self-managed teams led by senior consultants, and cost savings.

Key Challenges

Implement real-time anti-fraud monitoring for payment transactions
Integrate multiple payment transaction sources into a centralized big-data platform
Build analytical models to detect fraudulent or suspicious transactions
Integrate models into an online transaction monitoring front-end

Project Deliverables

Data integration pipelines for payment transactions
Hadoop-based data mart
Analytical fraud detection models
Integration of models into online transaction monitoring front-end
Governance and analytical model lifecycle management

Project Solution

Profinit provided consulting and development expertise in Hadoop technologies, proposing a 5-person expert team. They prepared data integration for payment transaction sources, collected relevant data in a Hadoop-based data mart, developed analytical fraud-detection models using Spark and Python, and integrated those models into an online transaction monitoring front-end. Profinit also assisted with overall governance and analytical model lifecycle management.

Project Outcome

Significant decrease of fraudulent transactions
Improved client NPS ratio
Effective workflow with self-managed teams led by senior consultants and architects
Cost savings compared to original estimation

Platforms

Tech Stack

Apache Hadoop
Apache Spark
Python

Client Endorsement

Overall Review Rating

Timeliness

Cost Rating

Willing to Refer

Quality of Deliverables

“The banking institution saw a significant decrease in fraudulent transactions and an improved client NPS ratio. Profinit's workflow was absolutely effective. They worked truly well as self-managed teams led by senior consultants and architects. Their services were also highly cost-effective.”
Martin Gerneš
Tribe lead
Date: 11/16/2022