Yuva AI

Computer vision AI systems such as self-driving cars and automated cancer detection are posed to transform our world. However, anyone trying to build these systems - be they governments, universities, private organisations and individuals - face the issue of obtaining high-quality and representative data to build these applications. Yuva AI removes the data barrier completely for anyone trying to build computer vision AI. We democratise access to data by giving engineers/data scientists a tool to generate data that reflects all populations in the world. Developers of AI can use our tool to then effectively produce systems that work for all races and that were unable to be built prior. This can impact hundreds of millions globally.

Computer vision, or the ability of artificially intelligent systems to “see” like humans, has been a subject of rigorous research and is now trying to be put into production across the globe. Applications of this technology such as self-driving cars and facial recognition are posed to impacting hundreds of millions of people across the globe but rely on millions of images of people to be developed to an accurate level. However, obtaining real footage of people and other objects is incredibly difficult. Firstly, privacy laws restrict access to data, especially in healthcare where HIPAA as well as other regulations prevent access. Of the data that is obtained, this must also be manually cleaned as well which provides an additional barrier and can take months to complete. As such, companies and data scientists building these systems have limited pools of data to access to build these systems which results in biased, non-performant algorithms in the real world.

Our solution is a development tool that any data scientist or product team uses to generate synthetic image data. This image data can be generated specifically to overcome the gaps in biased datasets available currently. For example, a user of our tool can specify that they want imagery of a middle aged African American male, and our platform generates the necessary imagery and accompanying files required. The user then feeds this data back into their AI system and the system can then identify African American males in the real world when in the past it couldn't.

Our tool serves any individual or organisation that is developing computer vision AI systems. We are working with engineers from organisations in the fortune 50, Max Kelsen, Genesis Care (Australia's largest skin cancer clinic) among other publicly listed organisations to help them improve the methods they are using to develop AI systems and to ensure bias does not exist. Engineers currently use extremely manual and painful methods to obtain data. They often travel to certain locations to take photos or videos. They may scrape the web but only find a dearth of useful imagery. They then must clean all those images and manually add metadata to them after ensuring they are compliant with privacy laws. Given how limited in scope these methods are, AI developed by these organisations is quite biased. By removing these barriers for them to obtain millions of images that are representative of all populations, they are able to generate solutions that can effectively work for all people instead of a sub section of them. By our estimates, they are able to develop AI algorithms 10X faster that are ready to get to market.

In this way, we are able to maximise the impact that artificial intelligence has on communities across the globe. Organizations can use our tool to impact communities globally in industries ranging from construction to defence.

My tool actively minimises human and algorithmic biases in developing artificial intelligence systems by tackling the problem at the source: the data. The data used to develop these algorithms are what makes these systems biased. Further, it is extremely difficult to obtain real data to develop algorithms. By giving the makers of this technology (the engineers, the developers and managers) a simplified and easy way to generate data on all populations, the biases in these artificial intelligence systems can effectively be eliminated.

I am currently piloting the software with numerous organisations. I have proven the technology works internally after conducting independent research on its effectiveness. I am now looking to obtaining third party validation from companies that need this solution to develop their algorithms. I am currently conducting pilots with companies in transport and robotics. This involves working closely with engineers and data scientists to test the product, educate them on the best ways on using it and how to evaluate its effectiveness. I primarily work with head engineers within firms who manage teams ranging from 10 to 15 people in size.

Yuva AI completely changes the way which organizations obtain data to build computer visionAI. Instead of using real images and videos, companies use our product to instead generate synthetic images and videos. These synthetic images have the relevant information to "teach" the AI robot everything it needs to know to function in the real world. Yuva AI's product streamlines the data process for a company and enables them to build AI 10X faster than traditional mean and reduce bias.

This is because our data doesn't have any privacy issues since it is synthetic and doesn't infringe on any privacy laws. This means companies don't have to go through the painful process of getting access to real data which is near impossible in many industries. A human also doesn't need to go through the images and manually label them like in this video () - it is all done automatically with our system.

I view this product as a catalyst for the computer vision AI industry. Where once applications could not be developed due to data constraints, they now can be. I have seen companies not be able to develop systems for skin cancer detection that could potentially save lives due to lack of data - now it is possible with synthetic data.

Current Number of Companies using product: 2. They are building applications that can impact 1 million lives in Australia.

In one year, the number of companies using our product is ideally at 20. At this point, this can directly impact 20 million lives globally.

In five years, I would like to have approximately 1700 companies using the product. By this point, the applications developed by these organisations should reach 100million people at minimum around the globe. These applications will range from healthcare to transport,

A simple indicator that is a proxy for how effective the solution is how many times the API is called every month. The more times the API is called, the more data is being generated and this translates to greater AI development. 

Further, the more companies that use the product, the more people are being impacted generally by the AI that is created as a result of the data generated by Yuva's platform.

Full time: Yash, the founder of Yuva AI

Contractors: Five contractors hired for short term projects over the course of a year.

Advisors: Three advisors who spend 15 hours with me between them every week discussing business and technical related issues (I ask a lot of questions).

For the five months from May to September last year, Yash cultivated intimate relationships with companies building AI by managing their data processes with an initial white glove service. This included Johnson & Johnson (AU, NZ & Japan subsidiaries) and Genesis Care ($1B+ valued company). From this, he has first-hand knowledge of the practical impediments in developing AI in critical industries like healthcare. After managing 250 people teams of labellers and witnessing the complexity of creating high quality datasets, he found that there was an enormous amount of value that we could deliver to these firms with an automated and scalable approach. His ability to achieve high SLA's and his access to clients such as Johnson & Johnson and Genesis Care puts us in a unique position to cater to this segment of the market as well.

In terms of gender, I am aiming to bring on Fei Fei Li, one of my favourite people as our first female advisor for the leadership team in the coming months.

As a leader, I have a responsibility to foster a company culture that is accepting and inclusive of all employees. I educate my team on what is acceptable and inclusive behavior and have strict policies in place for compliance with our diversity and inclusion initiatives. Creating a workplace where every employee feels safe and accepted is essential for a successful business.