Ilyes Kacher is a data scientist at autoRetouch, an AI-powered platform for mass editing of product images online.
I am a native French data scientist who started out as a computer vision research engineer in Japan and later in my home country. However, I am writing from an unlikely machine vision center: Stuttgart, Germany.
But I’m not working on German automotive technology, as you might expect. Instead, I found an incredible opportunity in the middle of the pandemic in one of the most unexpected places: an image-editing startup focused on e-commerce and powered by artificial intelligence in Stuttgart that focused on automating the digital imaging process. on all retail products.
My experience in Japan taught me the difficulty of moving to a foreign country for work. In Japan, it may often be necessary to have an entry point with a professional network. However, Europe has an advantage here thanks to its many accessible cities. Cities like Paris, London, and Berlin often offer a variety of job opportunities and are known as centers for some specialties.
While there has been an increase in completely remote jobs thanks to the pandemic, expanding the scope of your job search will provide more opportunities that match your interests.
Look for value in unlikely places, like retail
I am working on technology derived from a luxury retailer and I am applying my expertise to product images. Approached from the point of view of a data scientist, I immediately recognized the value of a novel application for a very large and established industry such as retail.
Europe has some of the most storied retail brands in the world, especially in clothing and footwear. That rich experience provides the opportunity to work with billions of products and trillions of dollars in revenue to which imaging technology can be applied. The advantage of retail companies is a constant stream of images to process that provides a playing field for generating revenue and possibly making an artificial intelligence company profitable.
Another potential avenue to explore is independent divisions typically within an R&D department. I found a significant number of AI startups working in a segment that is not profitable, simply due to the cost of research and the resulting revenue from very specialized clients.
Companies with data are companies with revenue potential
I was particularly drawn to this startup due to the potential access to data. Data on its own is quite expensive, and several companies end up working with a finite set. Look for companies that are directly involved at the B2B or B2C level, especially retail or digital platforms that affect the front-end user interface.
Leveraging this customer engagement data benefits everyone. You can apply it for further research and development on other solutions within the category, and your company can work with other verticals to solve your pain points.
It also means that there is huge potential for revenue gains to more cross-sectional segments of an audience impacted by the brand. My advice is to look for companies with data already stored in a manageable system for easy access. Such a system will be beneficial for research and development.
The challenge is that many companies have not yet introduced such a system, or do not have someone with the skills to use it properly. If you find that a company is unwilling to share in-depth knowledge during the courtship process or has not implemented it, look for the opportunity to present data-centric offerings.
In Europe, the best bets are to create automation processes
I have a sweet spot for early stage companies that give you the opportunity to build core systems and processes. The company I work for was still in its infancy when I started, and I was working to create scalable technology for a specific industry. The questions the team was tasked with solving were already being resolved, but there were numerous processes that still needed to be put in place to solve a myriad of other problems.
Our year-long efforts to automate bulk image editing taught me that while the AI you are building is learning to run independently on multiple variables simultaneously (multiple images and workflows), it is developing technology that does what images do. established brands. I have not been able to do. In Europe, there are very few companies that do this and they are hungry for the talent that they can.
So don’t fear a little culture shock and take the leap.