09 sep 2022
AI is changing the world in many amazing ways. From national security to online shopping to navigation to song suggestions on Spotify to healthcare, AI is revolutionizing all aspects of modern life giving people superpowers.
As amazing as this tool is, it’s crucial to understand that a tool is only as good as its master. A lot of people can get hold of a chisel like Rodin’s, but very few could create the likes of The Thinker.
That’s why at iScanner, we always say that our greatest achievement is people—a team of ambitious professionals we managed to put together, each of whom has got something to bring to the table. We’ve worked very hard to build trust, respect, and productive relationships within the team.
Two years ago, we made a very important strategic decision to form our own Machine Learning team. Looking back, it’s been one of our best investments so far.
iScanner’s Machine Learning is a team of seven people, all brilliant Machine Learning/Deep Learning/Computer Vision Engineers. Anyone who understands more or less what machine learning is about would agree that it’s more of a mixture of art and science than a practical skill set like programming or engineering.
As with any creative team, our ML team has very few rigid rules. However, there is one that they follow at all times: the merit of the opinion always wins over the title of the person who holds it. This helps boost creativity, create an inclusive environment, and sustain the value that each of the team members has to add.
It’s quite common for companies that develop AI-based products to use free-to-use, pretrained machine learning models. It saves money and time, and teams can concoct a solution really fast. However, iScanner’s Machine Learning Team has made a strategic decision not to take this easier route, whenever possible. Collecting enough data to train a machine learning model from scratch is a huge challenge. This, of course, means more hard work for the team, and more expense for the company in general. But considering that it helps us perfectly fine-tune our apps’ features for users’ needs and yields top-quality results, our ML team and company management to boot think it’s worthwhile.
One of the keys to successfully turning a brilliant idea into an attainable solution is high-quality research. This was the case with our new Scan Straightener feature as well. It took a lot of research work to get a deep understanding of user needs, consolidate data about different types of distortions, group those into categories, and determine the deliverables before we could even look into machine learning model design.
Distortion removal has proved to be a very complex and rather diverse problem to solve at a machine learning level. As challenging as it was, it gave us a lot to think about and inspired us. It’s quite possible that some of the algorithms will grow to be separate features or even apps.
At iScanner, we often like to say that we love what we do. “We love what we do” has even become our slogan. But for mobile app developers, simply loving what you do is not enough.
Mobile apps development is a unique niche in many ways. In quite a lot of industries, if you make a mediocre product by being frugal with research and technology but put it in an attractive wrapper, some people might still buy it for the looks. But when it comes to mobile apps, not only do you need an amazing idea, but you’ve also got to create an engaging design and smooth customer experience. No shortcuts here—it has to offer value for the customer.
We think the secret of our success is that we love our customers as much as our work. The Machine Learning Team is no exception. When asked what motivates them and gives them energy, most people said it was customer feedback.
By the way, the iScanner team is hiring! If you are a machine learning specialist, like what we do, and share our team’s values, please don’t hesitate to contact us at [email protected].