By continuing to browse this website, you agree to our use of cookies. Learn more at the Privacy Policy page.

Xenoss CTO Discusses ML, AI, and Data Science on DOU Live

Posted on June 9, 2021
Xenoss CTO Discusses ML, AI, and Data Science on DOU Live

Xenoss CTO Vova Kyrychenko has recently taken part in an online panel dedicated to ML, AI, and data science. The panel was organized by DOU, Ukraine’s largest online tech media and community. His fellow participants were:

Here’s our recap of Vova’s take on machine learning challenges and regulations, tech education, the future of ML and data science, and more.

ML regulations: the privacy-vs-comfort dilemma

“I once read a sci-fi book depicting a futuristic society that had no confidentiality. In fact, people refused their privacy for the sake of comfort. 

All of the innovative systems that are currently developed provide an extremely high level of comfort. Let’s take a look at a credit card as an example. Owning one means we consent to allow the bank to possess as much information about ourselves as possible. Banks know almost everything about what we live and breathe, and we’re completely OK with this.

Now that ML developments are in full swing, the question is whether we’ll be able to achieve the level of privacy we’d had before ML caught on. It’s quite unlikely, to be honest.

At the moment, we, humans, are at a wild stage, meaning we build and crank out new systems powered by advanced technology while having no knowledge of how these systems will affect our life as we advance. Every mistake we make in this process will have consequences. While they’re insignificant, no one cares. However, if a mistake led to someone’s death, everyone would care.

That being said, I believe the attempt to regulate the ML sphere is the first step to our understanding of human rights and privacy in a new ML-driven reality. Privacy can be sacrificed for the sake of safety, as the latter is of higher importance to us. We need to strike a balance between our comfort and privacy, and regulations will help us do so.”

Where to start your journey as a ML developer

“A mathematician was once asked what type of math was worth doing in the modern world. His reply was – do the math that drives you the most.

Today, machine learning is both science and art. There’s a broad range of ML-related domains you can tap into and master. Some people will fall in love with text recognition or NLP, while others may prefer to deal with sentiment analysis and extraction. If you need to choose, try as many things as possible before making the final decision.”

What makes a good data scientist

“A PhD in Data Science doesn’t guarantee you’ll be a good specialist. The problem here is that many so-called data scientists live in a perfect world with perfect data that can be easily analyzed and applied. 

However, reality is usually a far cry from that perfect world. Instead, we deal with poorly representative and highly unbalanced data and models that can’t be applied in practice without additional workarounds and integrations. A good data scientist understands it and knows how to manage and derive value from such data.”

How ML affects software dev project revenue

“At Xenoss, our data science results and ML-based predictive models are translated directly into financial metrics. 

On the one hand, finding the most effective ML model for the client’s product can be quite challenging and may require many iterations of an experiment. In this case, the total cost of the project will increase. Thus, the revenue will differ depending on the model used and the time spent on finding one.

On the other hand, for Xenoss clients operating in the adtech market, choosing the most effective ML model is crucial for the revenue of their business, not just the project. Having the most effective adtech platform with the highest conversion rate for end-users, which becomes possible with ML, brings an enormous competitive advantage to our clients. Thus, which model we choose/develop and how we apply it directly impacts the client’s financial success.”

The future of ML

“Today, it’s becoming more available for us to use what was expensive a year ago. In the future, we’ll see more projects leveraging data and ML technologies such as computer vision. We’ll also see significant developments and innovations in hardware that will enable ML evolution and commoditization. Theoretical models will become practically available.”

To read a full transcript of the panel, please visit the DOU website.

About Xenoss

Xenoss is a software development сompany building complex big data, machine learning and high-load solutions. The software Xenoss delivers is the tech basis of multi-billion businesses and is being used by Nestlé, Adidas, Virgin, HSBC. The company is headquartered in New York and has R&D centers located in Kyiv and Kharkiv, Ukraine. Follow us on LinkedInFacebook and Twitter.