How This Mumbai-Based Startup Is Using Computer Vision To Provide Video KYC

The potential of IoT and AI can be said as limitless. The key to finding the potential is basically to find practical use-cases where businesses can rapidly realise the value of these AI and IOT heavy investments. 

At the present scenario, the whole world is moving towards a digitally connected and automated world revolving around smart gadgets, homes and cars to animated customer engagement outposts, chatbots, and personalisation of products and services. 

Regarding the practical use-case of AI and IoT, one such use-case is known as Video KYC. Video KYC with its embedded capabilities can offer many enhancements, such as geo-tagging, liveness detection, image recognition, fuzzy matching, computer vision and optical character recognition, neural networks, and scalable infrastructure- to deliver better accuracy at a fraction of the cost. 

Subscribe to our Newsletter

Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.

To know more about how Video KYC works and how AI and machine learning is helping businesses to revisit their customer engagement operations, Analytics India Magazine got in touch with the founders of Mumbai-based startup Think Analytics.

Founded in 2014 by Amit Das and Monish Salot, Think Analytics is an analytics solution and services company that built multiple solutions and invests in products with a potential of disrupting the mobile data and analytics ecosystem. 

Flagship Products

There are several flagship products that the company is working on. They are mentioned below:

  • Algo360- Algo360 is a mobile app-based alternate data SaaS product, which provides score and insights to manage credit risk. 
  • Kwik.ID- Kwik.ID is a digital and Video KYC (SAAS) solution for customer onboarding and instant identity verification. It provides a seamless, secure and instant method of conducting KYC.
  • Real.CV- Real.CV is an AI-enabled contextual OCR solution that builds repositories of unstructured document designs on the go for structured data creation.
  • Third Eye- Third Eye (beta) is a state-of-the-art real-time, no-ball detection product using computer vision.
  • FlowXpert- It is a low code application journey designer that helps technology and business teams collaborate on building applications, including complex rule engines that power real-time decisions at the customer front-end.
  • Customer360- It combines the proprietary automated data engineering framework (ADEF) with industry embeddings that help visualise a 360-degree view of a customer’s buying journey.
  • Athena- It is a framework for lenders to completely automate their MIS and Dashboard requirements across functions.
  • F5- F5 helps lenders leverage data from a variety of sources to waive field investigation/ personal discussion calls.
  • Supply360- Supply360 helps procurement organisations (with dependence on a large number of vendors) optimise the spends using AI in traditional optimisation models, without putting any business value at risk.

How Kwik.ID Is Different From Other In The Market

To this question, Das explained, “Speaking of Kwik.ID, it was one of the first products fully compliant with RBI’s V-CIP guidelines that came in January, and we believe that our product really turns customer onboarding into a very seamless process.”

He added, “Kwik.ID’s edge comes from its extremely modern and modular design that allows for continuous innovation, in addition to our traditional expertise in AI and computer vision. Also, Kwik.ID is designed to ensure that the Video KYC tool reaches even the farthest of rural areas with provisions for a low bandwidth mode and guided step-by-step processes.”

Use Of AI & ML at Think Analytics

According to Salot, amongst other things, the product Kwik.ID comes built-in with OCR for standard as well as non-standard documents, cross-document face match, liveness detection for protection against paper and video attacks, automated signature extraction, speech to text and more.

Salot also mentioned that Kwik.ID is packed with cutting edge artificial intelligence, computer vision, big data analytics, and smart workflow design features. He added, “Be it our Facematch and Liveliness feature, or data-driven analytics, each and every step in Video-KYC is artificial intelligence reliant. Techniques like automated data extraction, data monitoring and streamlining complex processes, all rely on AI.”

Core Tech Stack

Das stated that tools like WebRTC sit at the core and help in powering the video and audio transmission. The company has built its own custom negotiation and connectivity processes on top of WebRTC to improve connectivity and ensure dynamic video quality change with changing network conditions of Kwik.ID.

All the portals and customer side interfaces are PWAs built with modern ReactJS+Redux architecture. The developers use a complex set of schedulers, distributed processing frameworks like Celery for Python and queueing systems like RabbitMQ coupled with ultra-low latency in-memory storage systems clusters backed by Redis.

The backend APIs are written in Python, NodeJS and GoLang with some services in Erlang. Also, The entire set of server-side components is 100% cloud/infrastructure agnostic and runs as a self-healing Kubernetes cluster. 

The system can switch over to using cloud-powered fully managed services like SQS, S3, DynamoDB, Lambda for AWS and their respective counterparts on Azure and GCP. Also, the entire data churning exercise takes place in an Elasticsearch cluster with Kibana used for visualisation and FluentD as the data and logs transport.

Salot added, “For the AI and ML parts of the product, our core stack is majorly Python backed with some parts written in Javascript to allow us to run models on client devices directly. We rely heavily on Tensorflow, PyTorch, Keras, OpenCV, Numpy, Numba for most of our activities with a distributed GPU cluster in place for rapid training and testing on demand.”

Hiring at Think Analytics

Currently, the company has an 80+ member team. Asking about hiring, Salot mentioned that the company looks for SWANs, i.e. Smart candidate can Work hard, is Analytically inclined, and (most importantly) is a Nice person to work with. 

He added, “We hire across channels, and we wouldn’t say that we have completely solved the talent crunch. We test rigorously, we train rigorously, but most importantly, we expect a lot from our people as they start working, very quickly. That learning curve with some pads, and some support – that’s what works.”

Future Roadmap 

“Each of the business lines has its own growth and expansion plans. We strongly believe that with the traction and product-market fit for both Algo360 and Kwik ID, we can consistently grow both of them for more than 200% YoY for at least 3-5 years on the back of new feature developments / addressing new use cases and international expansion,” said Das on a concluding note.

Ambika Choudhury
A Technical Journalist who loves writing about Machine Learning and Artificial Intelligence. A lover of music, writing and learning something out of the box.

Download our Mobile App

MachineHack | AI Hackathons, Coding & Learning

Host Hackathons & Recruit Great Data Talent!

AIMResearch Pioneering advanced AI market research

With a decade of experience under our belt, we are transforming how businesses use AI & data-driven insights to succeed.

The Gold Standard for Recognizing Excellence in Data Science and Tech Workplaces

With Best Firm Certification, you can effortlessly delve into the minds of your employees, unveil invaluable perspectives, and gain distinguished acclaim for fostering an exceptional company culture.

AIM Leaders Council

World’s Biggest Community Exclusively For Senior Executives In Data Science And Analytics.

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox