Cutting-edge innovation in computer vision and AI

Chennai based startup DeepQuanty Artificial Intelligence Labs is doing interesting work in deep learning AI ecosystem, combining human intelligence and machine learning to bring smart automation solutions for the BFSI and other sectors

DeepQuanty Artificial Intelligence Labs applies computer vision and artificial intelligence to solve problems in various sectors including BFSI, retail, e-commerce, and healthcare. The founders of the Chennai headquartered startup are Dr. Jayaram K Iyer, Mahadevan Jayaram and Sundar K. All the three of them have other successful businesses in the analytics and technology domains.

Mahadevan Jayaram, Co-Founder, DeepQuanty

Their products extract data from images using computer vision and process the data using machine learning techniques for better end-use. “Our mission is to drastically improve customer experience (CX) while improving cost efficiencies,” says Mahadevan Jayaram, Co-Founder, DeepQuanty.

Currently focusing on the BFSI sector, DeepQuanty later aims to leverage the patent-pending technologies in other domains such as healthcare, retail and such. The startup is self-funded as of now and will continue to be the same in the near foreseeable future.

With the current set of customers being in the BFSI domain, Jayaram explains, “BFSI is a heavy paper processing industry. Documents processed are cheques, bank statements, application forms, KYC documents. Our products aim to improve customer experience by way of reducing throughput time, increasing accuracy and lowering costs,” he says. For example, the recently introduced product SnapChek – an automatic processing engine, powered by AI, processes both printed and handwritten cheques. It extracts date to check for date validity, the amount in words and amount in figures and compares them, also extracts the account number, pings the core banking system for card signature, and compares with the extracted cheque signatures.

ZapSkore processes bank statements that are part of a loan application form. “It extracts all the in and out flow of cash, groups them into meaningful piles such as incomes and expenses, detects cheque defaults and so on. Such information is critical for credit risk scoring,” adds Jayaram.

The other products are EazyForm that extracts data from application forms such as savings / current account, credit card application, Loan application, and so on. While EazyKYC extracts data from most of the RBI approved KYC docs such as Aadhar, PAN, driving licence and passport.

Jayaram points out that innovation is at the heart of all products developed at DeepQuanty, “We believe there are two key ingredients that go into creating an organisation that fosters and encourages innovation – the people forming the team and the environment (company culture, innovation methodologies, etc.) they operate in.”

The hiring process is very stringent, and a talented team player is preferred over a super brilliant individual performer. “People are encouraged to explore and test ideas. We don’t condemn failures so long as the same mistakes aren’t repeated. Everyone in the team is allowed to bring a new idea to the table without fear of ridicule. We also expose every single member of the team to the customer production environment and adapt the product to perform in such environments. This brings everyone closer to the customer driving innovation that creates value,” he affirms.

Reminding that deep learning space, especially in the area of AI lead computer vision is constantly evolving, Jayaram states, “Newer and newer technologies are emerging that helps alleviate some of the most festering problems such as lack of large training data, processing capabilities, analytical techniques and launching applications in the production environment. Innovation is called for in each domain.” For instance, while creating SnapChek, an automatic cheque processor, the team had to create thousands of mock cheques from people for the AI systems to train. Printed documents were generated synthetically. “Processing capabilities required the latest and powerful Graphical Processing Units; however, it called for specific techniques to obviate the need for GPUs at the client environment. Banks are loathe to use cloud-based solutions and almost all the computer vision products call for cloud-based servicing. DeepQuanty developed technologies that obviate this need,” informs Jayaram.

The future plans include taking DeepQuanty to develop technologies that can be leveraged in multiple industry domains. “The current focus is the BFSI and therein processing documents. With some adaptations, the same technologies can be leveraged to solve problems in the retail and healthcare space,” says Jayaram, giving the example of cheque signature matching techniques which can be used to match two images on the internet thereby performing an image-based search.

Artifical IntelligenceComputer VisionDeep LearningDeepQuantyGPUmachine-learning
Comments (1)
Add Comment
  • Fatima

    Great article!