For this week’s startup column, Analytics India Magazine spoke to Lalit Mehta, Co-founder & CEO of Decimal Technologies to understand how the Gurugram-based fintech firm is leveraging emerging technologies to digitise lending.
Founded in 2009 by Lalit Mehta and Arvind Nahata, Decimal Technologies is a fintech startup that offers various mobile-based applications solving business requirements such as lead management, activity management and more. So far, the company has garnered over 200 thousand application users and disbursed more than $100 million in loans. Besides Gurugram, Decimal Technologies has offices in Delhi, New York and Singapore.
“Our aim is to empower the new generation mobile workforce and reduce customer acquisition costs by moving the digital touchpoint closer to the customer and making the entire onboarding process paperless and convenient,” said Mehta.
The company offers solutions in three key areas – Digital Lending, Customer Onboarding, and Incentive & Payout Management. Mehta said, “Digital lending is a focus area for us and through our digital lending platform, we offer a completely paperless, secure, API-driven omnichannel support to carry out all loan-related processes for banks and credit institutions.”
Vahana is a unified platform and an online API marketplace. Using Vahana customers can integrate services like digital KYC, video KYC, Aadhaar-based digital signatures and more to their platforms.
The company has recently launched a cloud-based lending marketplace, Saarathi. The product enables the digital transformation of loan files in under 30 minutes and can be accessed from anywhere.
What’s The Differentiator
“Saarathi is built on our proprietary No-Code Low-Code platform which facilitates scale with security and differentiates it from other products in the market. The AI-ML components it is built on help with seamless customer and lender matching without any manual restrictions through recommendations. We are striving to digitise the entire offline sources currently capturing 92% of the formal lending market in India with this product.
Saarathi is, in fact, a cost-effective platform and it benefits the customers by making approvals 5 times faster than the industry standard. For the next few months, our complete focus is on nurturing this product in the market,” said Mehta.
Use of AI @ Decimal Technologies
Decimal Technologies deals with use cases related to KYC, document recognition and information extraction, matching faces, images, signatures etc, developing and utilising alternate data sources for credit assessment, offer development, cross-sell and up-sell opportunity creation etc.
According to Mehta, the machine learning techniques the company uses range from traditional machine learning methods to deep learning applications. Decimal Technologies uses natural language processing (NLP), computer vision techniques, deep learning techniques such as Faster RCNN, Siamese net, among others.
Decimal has developed high-grade OCR mechanisms for reading a variety of KYC and other documents such as cash flow, balance sheets and P&Ls. For document detection and extraction, Decimal relies heavily on SSD/ faster RCNN based methods. For face matching, Decimal uses Siamese net-based face match. Decimal has also made matching algorithms for the deduplication of names and addresses across documents.
“Collaborative filtering forms the basis of a powerful recommendation engine that powers our lending platform while the appropriate allocation of tasks and leads is taken by a combination of linear programming and business rules. Also, one of the most powerful manners that Decimal uses its AI engines is by converting them as microservices and APIs. This allows the majority of algorithms to be applied across use cases,” he said.
Core Tech Stack
Decimal Technologies follow six principles that help the company to choose the right technology stack and architecture. The principles include scalability, interoperability, security, popularity, credibility and innovation. The core tech stack includes:
- Development: Java with Spring Framework. It is safe, portable, maintainable, scalable, and often the default choice in the market for many upcoming scientific applications such as natural language processing.
- Front End Web Development: Flutter has been the first preference for Mobile and PWA development.
- Library Selection: Libraries include Prometheus & Grafana for monitoring, Kafka, Elasticsearch, Logstash for Data Logging and Metabase for Reporting
- Data Store: SQL and NoSQL support
- Data Analysis: Various ML libraries like Sklearn, Pytesseract, Opencv, Keras, TensorFlow, Pytorch, etc.
- Testing: Jenkins
- Deployment: CI/CD (Continuous Integration/ Continuous Deployment) using best-in-class: JIRA, Jenkins, Docker, Kubernetes, Slack.
“We want to scale up and add a gamut of products. While the current focus for Saarathi is digital lending, we plan to expand it to include various kinds of loans, credit cards, insurance, etc in our bouquet of offerings. We also want to develop a lead generation platform for our customers, blockchain for invoice generation, third party integrations and much more in our mobile application itself. Meanwhile, we want to build a direct loan origination system with the bank, minimising the efforts of our lenders to check and approve the loan applications. Now, with Saarathi launched in the market, we want to expand our reach to any bank, financial institution and/or NBFC. In fact, anyone who aspires to be a financial advisor can associate with Saarathi and work with us in promoting financial inclusion via digital lending. We are expecting to build a $100Mn+ annual revenue business in the next five years, and for this, we are bringing extra pairs of hands and are aggressively hiring,” said Mehta.
Join Our Discord Server. Be part of an engaging online community. Join Here.
Subscribe to our NewsletterGet the latest updates and relevant offers by sharing your email.
A Technical Journalist who loves writing about Machine Learning and Artificial Intelligence. A lover of music, writing and learning something out of the box.