November 20, 2023
12 min read
Pioneering AI and Data Science in Ukraine
Maryna Hornieieva
Content Writer, Lviv IT Cluster

MindCraft.ai is a leader in Machine Learning and Artificial Intelligence development in Ukraine. Proven Data Science professionals at MindCraft.ai have already delivered over 17 projects of various complexity for startups and enterprise-level clients, including Fortune 500 companies.

We talked to Nazar Savchenko, Co-founder and Operations Director of MindCraft.ai, about business processes, the usage of AI in the ongoing war, the AI services market in Ukraine, and what the future holds for the world with the development of AI.

We founded the company in 2017. Initially, we were a team of six, but now we have grown to 14 members. We worked in completely different positions in outsourcing: as developers and architects. We met, started communicating a lot, realized each other’s strengths, and decided to focus exclusively on data science. Back then, there were a few specialists in this field, and only a handful of companies positioned themselves this way because the market was still relatively unknown.

Among our employees, there are individuals who, along with their parents, pioneered the development of the first neural networks in the 1980s. They are true enthusiasts. Our growth is organic, ranging from more to fewer team members, up to 16, depending on the tasks at hand. We define ourselves as a consulting company. We view our team members as consultants and engage them for specific periods of collaboration.

Eight out of ten of our clients face a similar issue: they require data refinement.

Remote office from inception

When we established the company, we realized that we didn’t need an office. For instance, commuting to and from the office in Lviv would consume 2 hours, along with associated expenses. During that period, our specialists were located throughout the country, these days they are based in Lviv, Odesa, Kyiv, Kharkiv, and even Canada. We didn’t just go after any project; we only accepted ones where we knew we could really help the client.

When we receive a project, we promptly provide the client with cost and time estimates. We believe in taking our time to gather data before committing to a project. This approach is highly appreciated by our clients, particularly those from Europe and the United States. There have been instances where we spent 3-4 months on a project, only to discover that the client’s idea couldn’t be realized or was significantly more costly than initially thought. 

Diverse client requests 

Our primary market consists of European countries, including Germany, Switzerland, and Denmark. We have a deep understanding of these markets as they share a similar mindset with us. Many German companies approach us based on recommendations from our existing clients.

 

Generally, clients come to us with two types of requests. The first one, although rare and quite costly, still attracts clients. It involves researching something entirely novel – creating something that doesn’t currently exist on the market, some kind of know-how.

Roughly eighty percent of our requests involve automating specific processes and tasks that people currently perform manually. The general belief is that AI can automate tasks within human capabilities. Initially, we conduct a research stage, which can last from one to three months. Following confirmation of the hypothesis and feasibility, we proceed with system development. This allows us to create comprehensive products that automate particular processes, reduce costs, and enhance overall business efficiency.

 

AI in defense

I can theoretically describe how AI can assist on the frontline: for instance, in recognizing specific objects. It could be a tank in a trench or anything else, identifying specific terrain features, regardless of the season, whether it’s winter, summer, in a forest, or a field. AI can also aid in navigation and even achieve precise aiming if the target is known.

The operator controlling it must detect these objects promptly to avoid detection. They don’t have much time to watch the video on their phone or computer, download it, or stream it online, as they are focused on the drone’s flight and locating the target area. However, the system is emotionless, which allows it to accomplish this task ten times faster. Consequently, it can be a lifesaver. Instead of spending a couple of hours, the operator can complete this task in just ten minutes, having all the necessary information at their disposal to save or protect lives. Just imagine a drone flying and identifying a specific target, such as a KAMAZ truck or a tank. It must decide which target to prioritize for an attack. It’s much more effective to target the tank than the KAMAZ. We can train the system to recognize these targets and assign priorities accordingly.

For example, when drones are in flight and transmitting information, it can be challenging to keep them in a specific position. They may hover over a specific target at an altitude of two hundred meters without any movement. Achieving this stability is quite complex because it involves calculating the wind’s direction, whether it blows to the left or right. The drone needs to be able to make these adjustments independently to remain in the same location within a specific square with minimal movement. These are also essential nuances, and, in principle, we are capable of managing these aspects as well. We can work on it.

Our first objective is to establish multiple data science teams across Ukraine.
A closer look at modern warfare

I think Russians have a higher level of AI utilization than we do. But this is just my opinion; They probably have two stages. The old stage involves Soviet H-55 missiles that they currently use against us. These developments date back to the 1970s-1980s. You could say these were their initial attempts at using AI.

Now they’re in a new stage. I believe they’ve advanced because they prepared for war for a long time. They never restricted themselves, and they didn’t adhere to any moral norms. I think in the last 10-15 years, they’ve been working diligently in this direction. They’ve invested a lot into it and continue to do so. I must admit that they have quite intelligent data scientists. I don’t think all of them have left the country.

Their Shaheds can be directed toward a specific area, following a certain trajectory. These drones consider the terrain, including mountains, and determine whether it’s better to descend or ascend. Consequently, they can autonomously select the correct trajectory or modify it.

 

How can you intercept or shoot down a Shahed? The challenge lies in their evasive nature. They can also alter their trajectory; one moment, a drone may be flying west and then suddenly make a sharp turn to the south.

We need to detect them promptly and deploy mobile systems capable of shooting them down. So, in terms of automation methods, we need to develop or acquire more of these systems. The best course of action would be to destroy the facility in Russia where the Shaheds are manufactured. Instead of dealing with the consequences, we should eliminate the root cause. Certainly, they possess substantial resources in the war, far more than we do. However, we employ our resources judiciously. The trend indicates that in two years, we have grown stronger than Russia, which had been preparing for this war for decades.

A closer look at modern warfare

I think Russians have a higher level of AI utilization than we do. But this is just my opinion; They probably have two stages. The old stage involves Soviet H-55 missiles that they currently use against us. These developments date back to the 1970s-1980s. You could say these were their initial attempts at using AI.

Now they’re in a new stage. I believe they’ve advanced because they prepared for war for a long time. They never restricted themselves, and they didn’t adhere to any moral norms. I think in the last 10-15 years, they’ve been working diligently in this direction. They’ve invested a lot into it and continue to do so. I must admit that they have quite intelligent data scientists. I don’t think all of them have left the country.

Their Shaheds can be directed toward a specific area, following a certain trajectory. These drones consider the terrain, including mountains, and determine whether it’s better to descend or ascend. Consequently, they can autonomously select the correct trajectory or modify it.

 

How can you intercept or shoot down a Shahed? The challenge lies in their evasive nature. They can also alter their trajectory; one moment, a drone may be flying west and then suddenly make a sharp turn to the south.

We need to detect them promptly and deploy mobile systems capable of shooting them down. So, in terms of automation methods, we need to develop or acquire more of these systems. The best course of action would be to destroy the facility in Russia where the Shaheds are manufactured. Instead of dealing with the consequences, we should eliminate the root cause. Certainly, they possess substantial resources in the war, far more than we do. However, we employ our resources judiciously. The trend indicates that in two years, we have grown stronger than Russia, which had been preparing for this war for decades.

Involving more data scientists in defense projects

We are not referring to data scientists as individuals. When there is a team of five or six data scientists, they collectively brainstorm ideas, provide constructive criticism, collaborate on development, and work toward a common goal. They are much more effective as a team. Once they have access to the data, their enthusiasm will be rejuvenated, and they will be capable of automating various processes. Additionally, grants should be available for drone manufacturers.

The government has recently simplified regulations, allowing for the production of various types of unmanned aerial vehicles in Ukraine. Furthermore, the government is prepared to invest in this sector by placing orders with Ukrainian manufacturers. These developments may take anywhere from one month to six months or even a year. Nonetheless, they remain highly significant because we cannot predict what will happen in a year. But some of these developments have the potential to save numerous lives within that timeframe. We still require research to improve navigation and target selection. Drone manufacturers are already offering grants, specifically approaching data scientists and stating, “We have a drone, and we need a system developed for it to assist in selecting the right target or determining the optimal route.”

Our company currently exclusively participates in defense projects on a voluntary basis. We are motivated by the desire to see our ideas, specific developments, or hypotheses realized, even if not by our own hands, then by others. We willingly share our knowledge because we firmly believe that we are all working towards the same goal. We share a common vision: to protect our loved ones and our country.

Looking into the future

Certainly, our first objective is to establish multiple data science teams across Ukraine. Currently, we have just one person in Kharkiv, and two or one in Dnipro. Our aim is to establish local teams in every city. Additionally, our goal is to demonstrate that Ukrainian data science consulting is thriving and operates at a high level. Our professionals are highly intelligent and capable of effectively managing European and American businesses, even despite the war. Our clients who have continued working with us express their satisfaction.

Therefore, despite the challenges such as power outages, shelling, and whatnot we are committed to enhancing our service delivery. We now fully grasp the price and the importance of the issue. And the fact that even in times of war, we provide only high-quality services, is a very good indicator.

Read more in the full version of the magazine issue
Read issue #20
Copied!