Case study · Digital lending

Modernizing Purple Finance’s loan analytics: 5X faster insights with real-time dashboards and automated ETL

Algoscale replaced commercial BI and semi-manual scripts with an open-source analytics stack on Apache Superset and AWS, and a risk dashboard that tracks defaults, fraud signals and compliance in real time.

Get the PDF Talk to our team

purplefinance.in ↗

60%decrease in manual data handling time

25%reduction in NPAs

5Xfaster loan approval TAT achieved

About the client

Purple Finance Limited is a digitally enabled secured lender serving micro and small enterprises in India that traditional lenders underserve.

It writes secured business loans against property, listed on the BSE under PURPLEFIN.

Sector: MSME secured lending
Country: India
Listing: BSE: PURPLEFIN

About the Company.

India’s digital lending market is projected to reach $150 billion by 2025, driven by fintech innovation and increasing access to credit across underserved segments, Riding this wave is our client - a leading Non-Banking Financial Company (NBFC) specializing in digital lending, known for managing large-scale loan portfolios across diverse borrower profiles.

With a strong focus on data-driven decision making, the client was already leveraging a mix of cloud-native and legacy systems, but their rapidly expanding operations demanded a more unified, a real-time analytics backbone to streamline loan processing and risk management,

60%decrease in manual data handling time, allowing teams to focus on strategic tasks.

25%reduction in NPAs enabled by real-time fraud and compliance monitoring.

5Xfaster loan approval TAT achieved, dropping from 24–36 hours to under 6 hours.

Solution Summary.

As a leading Data Consulting Company, Algoscale implemented an enterprise-grade, open-source analytics stack, anchored on Apache Superset and AWS-native services to modernise analytics and automate data workflows for the client. Our solution eliminated BI licensing overhead, while enabling real-time visibility into key business metrics like TAT, loan defaults, fraud risk, and regulatory compliance.

We built a metadata-driven ETL framework using AWS Glue to seamlessly transfer and transform loan data from AWS S3 to PostgreSQL, ensuring support for schema evolution, incremental loads, and deduplication logic.

Additionally, a custom RCPU Risk Dashboard was deployed to provide comprehensive insights into borrower behavior, early default signals, and operational bottlenecks - empowering the client to reduce NPAs and enhance customer experience.

Customer Challenges.

Despite being a digital-first NBFC, the client’s data landscape had become increasingly fragmented, creating critical bottlenecks in operations, risk management, and decision-making.

Scattered Loan Data Across Systems.
Loan information was distributed across LOS (Loan Original System), LMS (Loan Management System),and spreadsheets stored in AWS S3, making it difficult to unify and analyze data in real time.
Lack of Real-Time Business Intelligence.
Existing BI tools were either cost- prohibitive or lacked the flexibility needed for on-demand analytics. Stakeholders had limited visibility into loan performance, risk exposure, and operational metrics.
Fragmented TAT Tracking.
Turnaround Time for approvals was tracked manually or across siloed systems, resulting in delayed insights and slower loan processing.
Manual and Rigid ETL Processes.
Data movement from S3 to PostgreSQL relied on semi-manual scripts, lacking schema adaptability, incremental load handling, and deduplication - leading to inconsistent data pipelines, and provide real-time, actionable analytics without adding license overhead.

Algoscale Solution.

Open-Source BI Setup.
Deployed a self-hosted Apache Superset instance integrated with PostgreSQL to deliver interactive dashboards, eliminating expensive BI licensing costs.
Metadata-Driven ETL Pipeline.
Built an AWS Glue-based pipeline supporting schema evolution, incremental data loads, and deduplication using composite keys and hash validation.
Automated & Orchestrated Data Flow.
Automated data transfer from AWS S3 to PostgreSQL, using AWS Step Functions to handle errors and enable job retries for seamless orchestration.
Real-Time Risk Monitoring.
Developed a centralized RCPU Risk Dashboard powered by anomaly detection and rule-based logic to track loan defaults, fraud signals, and compliance metrics.
CI/CD Enabled Deployments.
Implemented CI/CD pipelines using Terraform and GitHub Actions to ensure consistent, automated deployments across ETL scripts and dashboards.

Algoscale Differentiators.

Open-Source First Approach.

Delivered an enterprise-grade analytics stack using Apache Superset, eliminating BI licensing costs without compromising on functionality or scalability.

Integrated Risk Intelligence.

Embedded fraud detection and compliance checks directly into dashboards, improving visibility and reducing NPAs.

Metadata-Driven ETL Architecture.

Designed a flexible ETL framework using AWS Glue that adapts to schema changes, supports incremental loads, and ensures clean and deduplicated data.

DevOps-Enabled Delivery.

Accelerated go-to-production cycles using Terraform, Docker, and GitHub Actions for seamless CI/CD deployment of both data and BI layers.

Real-Time Data Pipelines.

Enabled near real-time ingestion and transformation of loan data across S3, PostgreSQL, and visualization layers- improving responsiveness for operations and risk teams.

Values Delivered.

$40,000 in annual BI licensing costs saved by replacing commercial tools with Apache Superset.
5X faster loan approval TAT achieved, dropping from 24–36 hours to under 6 hours.
Improvement in data accuracy through automated ETL and schema-aware validation.
25% reduction in NPAs enabled by real-time fraud and compliance monitoring.
60% decrease in manual data handling time, allowing teams to focus on strategic tasks.