Jukshio has mastered the dual nature of verifying identities through optimizing end-user experiences with the risk appetites of businesses seamlessly and in a manner that can be controlled by the business in near real-time. In our 4 years of experience working in the identity verification business, we have had to adapt furiously to perhaps the sharpest learning curve in the industry as we went from zero to a billion transactions upon winning the identity verification chunk of business of one of our largest customers, a telecom giant. 

‍

Standing on the shoulders of giants 

‍

Our roots are grounded in strong academic underpinnings, as our founders believe in the power of education and opportunity to transform ecosystems. Team Jukshio has learned on the job having to rapidly evolve to the reality of fraud and the need for performance at scale, convenience, and simplicity in business processes. Entering a mature identity verification market, taught us that carving a unique niche for ourselves meant rapidly prototyping and shipping solutions to the client as regulations changed and business risk appetite fluctuated. Customizing solutions built on open-source code meant building credibility and support in developer communities. Our core belief in innovation centers around making that small dent in the identity verification framework to be able to support communities and individuals whose documents would not be accommodated by the computer vision models of the day. Jukshio, from day one was serious about addressing the risk of unintentionally excluding populations. We document challenges that genuine end users face with the status quo of face recognition (FR) and face liveness (FL) algorithms prevalent in the identity-proofing industry and how we refined our solution and trained our models to accommodate their specific conditions. 

‍

🚀 Jukshio’s solution performance is agnostic of the low network bandwidth of end users

‍

Through a combination of optimized non-lossy compressed image data formats, and reducing the number of touchpoints in the end user workflow, Jukshio is able to provide an uninterrupted experience to the end user through progressive loading and minimal to no use of non-functional components, so the end user can access and interact with the critical components of the solution. 

‍

🤳 Our models excel at handling documents photographed in the wild 

‍

Jukshio supports all government-issued ID cards in India. Our models are able to handle and process photos taken in low lighting conditions or overexposed images. We are able to iteratively train our proprietary models on patterns recognized in data, till photos taken in such conditions that are decipherable are no longer rejected by the models. We provide immediate feedback to the user on whether the image taken by them is suitable or not for the process. The Jukshio UX is non-fussy about only accepting that perfectly angled and composed shot and can tolerate a considerably lower degree of image fidelity. The SDK we provide has corrective error messages that are delivered instantly to the end customer who can immediately rectify their photos to click the perfect image of the ID document or a selfie. 

‍

🏋️ The Jukshio low-code platform can process worn-out and soiled IDs 

‍

Jukshio allows leeway for documents in different states of wear and tear. With the history of identity documents in India starting in colonial times and the ration card being around since the 1940s, we understand documents might not always be in the crisp, flawless state required by computer vision models, so team Jukshio improvised to train our models on identity documents that were fraying and in poor condition. 

‍

🚷 Precisely document, reject, and block or report repeat attempts at spoofing

‍

Identify repeat attempts at spoofing the system by a single end-user using our spoofing models. Spoofs that our models detect include digital spoof - where the user presents the image of an ID on a digital screen, paper spoof - in which the user covers a portion of the ID document with paper, photo pasted on ID documents, number spoof - physical or digital tampering of the ID number on the document and photoshopped identity documents. On average, our systems receive tens of thousands of spoofs a day, consisting mainly of photos pasted on ID documents and digital spoofs. 

‍

👨‍👩‍👧‍👦 Cultural sensitivity

‍

Respecting cultural norms and conventions is essential in identity verification. End-user convenience and comfort is paramount during the identity verification journey. Jukshio’s IDV solution supports several facial occlusions such as turbans, scarves, spectacles, and winter caps that don't obstruct the face. We support medical masks but not sunglasses. Sunglasses cover the eyes and eyes are critical to facial recognition algorithms. Married women in certain communities have to cover their faces as they adhere to multiple traditional customs and their faces are not completely visible when captured for the passive face liveness check. 

‍

🌏 Naming conventions in India vary widely across geographies and significantly differ from standard Western naming conventions. Women’s maiden names can be completely different from their married names, as usually the surname changes and in some cases, the first name changes too. Names can often be phonetically spelled on different identity documents, for example, an ‘ee’ in a name may be spelled as an ‘i’ in another document or a name containing an ‘oo’ might be spelled as ‘u’ elsewhere. Inadequate recognition of these naming practices can lead to misrepresentation, confusion, and a frustrating user experience. 

‍

A culturally inclusive identity verification solution not only ensures compliance with KYC requirements but also fosters trust, increases user adoption, and promotes a positive user experience. By embracing the diversity and cultural richness of India, institutions can create identity verification systems that empower all individuals, regardless of their cultural background or digital literacy level, to participate confidently in the digital age while upholding necessary security measures.

‍