Lukasz Kuncewicz, enigmapattern.com
Hi! We’re so happy to have you here! Could you briefly introduce yourself to our readers?
Hey, my name is Lukasz Kuncewicz. I’m Head of Data Science at Enigma Pattern where I have the pleasure of working with many talented data scientists. Together we solve various issues for our clients and help them get the most value from their data. Can’t think of a better job, really.
So, it’s true that when you choose a job you love, you’ll never have to work a day in your life?
Yes, in my case, absolutely!
That’s really awesome, Lukasz. And when did you get started with data science?
Well, thanks to my dad I started to write computer programs at the age of 7. That defined my life, I guess.
Wow! That’s fascinating! Sounds like data science runs in the family! You have quite the background, but when exactly was the first time you heard about the field?
The first time I touched Machine Learning was more than 20 years ago, and the attempt didn’t go so well… However, today the landscape is totally different: algorithms do work, we have more than enough data to feed them, as well as a proper infrastructure to run them on… Definitely exciting times for data scientists.
How did you end up working in data science?
I started a few years ago when I realized “OMG, this thing is actually working!”. I dropped everything and just devoted myself to improving my algorithmic skills in the fields of data manipulation, business analysis, and AI. It wasn’t an easy decision since I have a family. I am a father of two and a stable lifestyle is important… But my passion for data won.
Oh, nice, following your dreams once you have a family is not an easy choice to make. We’re sure our readers are just as impressed as we are right now. Lukasz, you are the Head of Data Science at Enigma Pattern. Could you tell us a bit more about the services you are offering, and how you create value for your clients?
In terms of work, we’re very pragmatic. We always try to be as practical as possible. Our main goal is to create a competitive advantage for our clients by using Artificial Intelligence. It can be anything from improving their products and internal processes to extending their knowledge about what is happening with their business now or will happen in the future, based on current data. If you believe the IT revolution is about automating processes and scaling them up, then the AI revolution is all about making products and services smart. And this is exactly what we do.
Fantastic! Now, Lukasz, could you share which industries the companies you typically work with operate in?
Well, we’re not really specialized in any particular industry. Most of our projects tend to focus on real products. We bring solutions that make them smart and compensate for their imperfections. We also improve the internal processes of companies. For example, we can predict churn among their clients and select the most urgent cases so they can act before it’s too late.
That’s quite the range of solutions you offer… Tell us, though, which is the area of improvement that is typically the easiest to address?
You know, there is so much confusion about what Machine Learning can and can’t do, and what it requires to work. If possible, we try to find a thing that is easy to understand and, at the same time actionable. We focus on an issue which doesn’t bring too much hassle for the client, and then just model it.
Sounds like a smart strategy. Still, is there something that allows you to achieve some quick-wins when you start working with a new company?
Well, to be more concrete, it could be something as easy as predicting fraudulent orders in your eCommerce shop, for instance. Once the model is done, we can calculate the amount the organization could have saved if they had used the model.
Why do you believe this makes a difference?
I think communication is the key. We don’t say “we can create a behavioral model of your clients’ activities”, but rather “you lost this money because you didn’t have our model”.
Yes, that will definitely convince a business-oriented person to sign up for Enigma Pattern’s services. Can you think of a situation when you have worked with a given organization and you felt especially proud of what you helped them achieve?
Sure! One of our clients is a Top 5 toy producer in the world. We’re currently working on ways to incorporate Artificial Intelligence into their products. I can only imagine how, in a few years’ time, my kids will be playing with these super-intelligent toys. I see myself saying “You know, it’s actually your daddy who made these toys so smart”. Can’t think of anything that could ever make me prouder.
And we can’t think of anything more heart-warming! Thank you for sharing this with us, Lukasz!
Now, in terms of day-to-day practices, which tools and software are essential for your team (coding languages, data management tools, visualization tools)?
To tell you the truth, it depends on the type of project.
For example, in R&D projects, our work is mostly Python-based, but Python is just a backbone. I mean, if it is quicker to implement an idea in PyTorch than in Python, we will switch to PyTorch. Or Caffe. Or TensorFlow. There are numerous possibilities. But still, we would get back to Python to gather results, run analyses, and so on.
And when it comes to production projects, it all depends on the client’s infrastructure. Anyhow, we usually work with Python, Java, Apache / AWS / Google Cloud infrastructure… Sometimes Unity.
That is really useful to our readers, thanks. Now, it’s time for the Tip of the interview. Is there a nifty tool that you discovered or were introduced to which you now can’t live without? Would you share it with aspiring data scientists?
Yes. That would be charts. No kidding!
Honestly, charts. So many times I see aspiring data scientists go into this “I -have- the- data- now- let’s -run- a -model -on- it” mode, before they even see the data. And I get it, sometimes it’s hard to visualize the data – but still, that is the first step. A lot of different charts that will tell you a lot about your data. (see chart types and how to select the right one)
Also, paying attention to where your model makes mistakes. It’s not enough to know, for example, that its accuracy is 80%. It is vital to understand where this 20 % come from, what is common between them, and how to change the training data set to compensate for this.
That was actually some pretty serious insight for new data scientists! Awesome. Lukasz, every time we conduct these interviews, we finish with some nerdery. What is the one nerdy thing you would like to share with the world? It doesn’t have to be data science related at all.
I am proud to be a nerd! I love Star Trek! In my life, I’ve had more laptops than girls I’ve dated, and I can’t tell if my shoes match my trousers. Go, nerds!
Yep - we’re proud to be nerds ourselves! Thank you on behalf of all cool Star Trek fans with mismatched shoes out there! Pun of the interview. We usually say goodbye to our readers with a joke. What’s yours?
Ok. Here it goes. Infinitely many mathematicians walk into a bar. The first says, "I'll have a beer." The second says, "I'll have half a beer." The third says, "I'll have a quarter of a beer." The barman pulls out just two beers. The mathematicians are all like, "That's all you're giving us? How drunk do you expect us to get on that?" The bartender says, "Come on guys. Know your limits."