Jio Case Study

Facial Recognition and Liveness technologies

Executive Summary 

Fraudulence in digital identity verification is rapidly growing due to ever-evolving methodologies. Service providers want to onboard as many customers as quickly as possible. But the sophisticated methods by some fraud users trick the system, jeopardize an organization's security, and attract hefty penalties when regulators audit such discrepancies. This report analyzes the issues faced by our client Reliance Jio and the technological innovations provided by Jukshio for KYC.

The Client and the bottlenecks

The client, Reliance Jio, faced issues with accuracy, cost-effectiveness, and the time taken to process KYC. The existing solution’s inability to handle sophisticated fraudulent attempts and the error rate during the processing of KYC concerned the client. Jio approached us for solutions as we were already engaged with them on their attendance management system with the deployment of FR+FL Solution named mHere.

TRAI regulations say no more than 9 SIMs from the same service provider. With a 500-800M active subscriber base at Jio, grouping people was nearly impossible. With a daily capacity of 1-5M onboarded daily, there was a requirement of 4000000000000000 decisions each day to scan out fraudulent documents.

Fraudulence involved agent and user-based attempts to fool the system. For instance, one user could use different ID documents across states. Due to demand, agent-involved fraudulence was also high. Agents tried to beat the system by using their photographs while not being physically present at KYC.

Jukshio’s Solution

Our AI learned from millions of samples. With AI, we built a document verification technology that can detect any kind of discrepancies in the document, like improper document templates, blurred documents, partial documents, and photocopies. The AI could also detect spoofed IDs wherein any details of the documents have been altered or tampered with. Our unique Human-Machine combo allowed the process accuracy to 99.998%—this differentiated Jukshio from other competitors in the market.

The team leveraged AI to make Facial Recognition and Liveness technologies highly accurate. They formulated a Document verification technology capable of extracting data and detecting any irregularities in the documents, such as spoofed or tampered documents. Constant human-aided machine learning helped the AI tackle the different types of fraudulent practices used to trick the system.

In user-involved fraudulence, we used our facial recognition technology to scan multiple formats of docs and created a non-name-based grouping of people. The facial recognition technology helped us to detect 567079 violations.

In Agent-involved fraudulence, we used facial liveness technology to identify agents responsible and blocklisted 5000 out of a total of 1 lakh. Our mechanism could sustain the pace and rate of blocking fraudulent agents. In a given month, 20,0000 instances of agent-involved fraud are detected.

The technology we provide is highly customizable. It allows the client to modify their business workflows as and when needed. Our unique human-aided Machine Learning helped us achieve an accuracy of 99.998%, one of the best in the industry.

Key highlights and impact we made

The client saved 91% on overall expenses and boarded over 445,072,767 users as of August 31st, 2022. Our solutions add value in terms of

  • Cost-effective and convenient technology that uplifts the revenue segment.
  • Regulatory audits found Zero errors.
  • Superior methods of fraud detection using AI-based auto modules and Manual re-verification.
  • Our solutions enable customer-centric customization that meets industry requirements and trends.

Our KYC audit led to Jio eliminating fraudulence in their business and in addition, they adopted our onboarding KYC solution to completely streamline their business.

No items found.
Published
Category