Today’s article is the solid proof that networking on conferences is really worth it. Joining InfoShare for the second time (you can read more about 2021 edition in my article from last year), I’ve decided to get out of my usual comfort zone a little and talk to as many people as possible. Also, my main motivation was collecting useful information about how other tech companies are structured regarding machine learning teams. I wanted to know what are the reflections of their employees about working for their employers, what is expected from data scientists/machine learning engineers in specific work places and how do specific employers define those roles. Also, what’s most important – how do enterprises deal with knowledge sharing and collaboration across data analytics teams in 2022?
To my great joy, after several good conversations, I came across Anna Brodecka, Product Owner for Silky Coders giving her speach of one of the projects. I have heard a lot good words about this company recently and now I finally understand why. Silky Coders’ stand was full of extremely enthusiastic, passionate and helpful people, willing to share their experiences with data analytics. “Why not to take advantage of this opportunity and continue the very nice conversation on the blog?” – I have thought. Fortunately Ania was positive about my idea and now I can invite you to read our discussion. What’s even more exciting, we have managed to encourage one more wonderful guest to the initiative – Kazimierz (Kazik) Najmajer who’s not only another passionate employee of Silky Coders, experienced data science practictioner but also my colleague from Mathematics lectures on Gdańsk University of Technology back in time. 🙂
Interview with Silky Coders
Sandra: Hello! Thank you so much for agreeing to answer a few questions. To start properly, could you please tell a few words about yourselves and your current employer? What kind of roles do you play in Silky Coders?
Kazik: I am a Senior Data Scientist. I am focused on inventing and deploying algorithms that optimize and automate processes in the company.
Ania: I am a Product Owner, interested in machine learning and AI for good. I have over 8 years of experience as a manager in R&D (Research & Development) projects in various industries. Personally, I consider it very important to connect the world of technology and business. I like learning about business issues, creating a vision of the project and taking responsibility for it. In Data Science department in our company, product owners are also the leaders for analysts teams, so it is also important for me to grow in this role and become a better leader.
Silky Coders is a company which has over 700 employees. We provide IT services, software solutions and data science for the whole LPP Group which has over 24 000 employees. LPP has 5 fashion brands – Reserved, Cropp, House, Mohito and Sinsay so we deal with data, algorithms and whole data science in the fashion industry.
Sandra: What kind of data sources do you deal with in Silky Coders?
Kazik: Mainly structured data – warehouses stock, sales. In general, various info on what is sold in our stores, on our logistics and ecommerce analytics. Sometimes we work with other kinds of that like in RFID project (radio frequency identification using as a electronic tags in our stores) where we have raw logs from tag reader or in other projects where images or geographical data might be used.
The aim of RFID project is to predict the location of an item (e.g. t-shirt) in a store using history of radio signal strength (RSSI). Data is collected by stores’ employees who take measurements with RFID readers during doing inventory. We use ML algorithms to assign items to fixtures.
Note: For Polish readers, if you’re interested more in RFID project, you can explore the topic using dedicated article on Silky Coders’ website here.
Ania: I agree with Kazik, at Silky Coders we have a huge amount of data concerning almost all business processes carried out in our company.
Sandra: How your customer (LPP) is collecting it? What about GDPR? How much information can we gather about our customers and how?
Ania: We do not collect or store any personal information that would be inconsistent with the GDPR, we work on aggregated data that gives us a big picture of the general trends of our clients behaviour.
Sandra: What do you think about tech world going fast forward and how Silky Coders suits this image?
Kazik: In my opinion we try to keep up with the changing world of technology. We use the cloud (GCP and Azure), and we are up to date with modern data stack. Of course, it takes much effort in the ever-changing world of data.
Ania: On the one hand, we try to keep up technologically, and on the other hand, we innovate ourselves and conduct research and development projects, such as money mapping, in which we have shown that we can locate style colors using RFID and algorithms developed by Kazik. We also conduct classes for mathematics students at the Gdańsk University of Technology, which we introduce to the world of practical application of data science in business.
Sandra: I have heard about your lectures there! Could you please tell more about it? What can students expect after completing your course/lecture? Do you have any advice for beginners just starting their career paths in data analytics? How to stay motivated? If only there was such cool machine learning oriented activities in Gdańsk University of Technology when I’ve studied there!
Kazik: We teach MSc in mathematics students on Gdansk Tech. Basically it is project base teaching focused mostly on practice so students can have a glimpse of how we really work. My advice is that one has to learn, do projects, courses. Talent/intelligence might be necessary but without hard work they mean nothing.
Ania: I will add that it is important not to be afraid to ask to gain knowledge, especially when you have the opportunity to work with practitioners. It is worth being brave and reliable at the same time and remembering why we provide solutions, always keep your head on what needs and what stakeholders our solutions are to respond to.
Sandra: What kind of projects do you work on in Silky Coders? What professions do you gather in the hub? Is it mostly focused on data science, machine learning and AI or do you also develop other software/products?
Kazik: Silky Coders as a whole is concerned with various kinds of IT. Basically whatever tech comes to your mind – we do it. But in the Data Science department, we are mostly focused on data science and machine learning.
Sandra: How many data science employees (data scientists, machine learning engineers, data engineers, etc.) do you have currently? How are the teams distributed?
Kazik: In Data Science we have around 100 (including Business Intelligence teams). Mostly data analysts and data engineers. We work in scrum teams consisted of data scientists, data analysts and data engineers.
Ania: We have 7 teams, and the support team consists of scrum masters, data science consultants and other specialists.
Sandra: Who is responsible for the deployment of a final application? Is it a skill that every data scientist should have or is it maintained in another way (e.g. MLOps or specific people)?
Kazik: We at Silky Coders are well aware of the necessity of having MLOps/ML engineers in our teams. Our data scientists and data engineers have in-depth knowledge and experience regarding the deployment and productionizing of machine learning models.
Ania: We share knowledge and competences in teams and we also prepare some developers to perform MLOps tasks.
Sandra: What is your approach towards knowledge sharing across your employees? Do you enable any kind of mentoring or internal trainings/reskilling or focus more on recruiting candidates with the right technical skillset from the market?
Kazik: We are very much focused on internal training. Many of our data scientists and data analysts came to work with us after graduating. Oftentimes in Mathematics where we taught them. Similarly, many data engineers are former data analysts. Thanks to that they have profound business knowledge. Definitely an asset in their job.
Ania: Sometimes we recruit employees with many years of experience, but more often we focus on the ability to learn and analytical skills, and above all, the enthusiasm that is needed to understand such a large business. As Kazik said, we also enable internal transitions – recently a great data analyst has become a great developer. Understanding our business needs is important in every role.
In Silky Coders we have programs that allows changing the position, but we also do it on our own, as in this case. It usually starts with earning new skills, charting a path with your supervisor, and then carrying out tasks in a new area. When the employee feels competent, he/she takes the exam and, having passed it, starts regular work in the new position.
Sandra: I have listened to your presentation during InfoShare and I already see how varied and fun can your projects in Silky Coders be. You don’t even need laptops at some stages, running around your stores’ aisles! Could you both please share your reflections about the coolest projects you have worked? Can be the ones done/ongoing for Silky Coders or from your previous experiences.
Kazik: I suppose the one you are mentioning, item location using RFID project was the most interesting. What is also important whatever we do, we have great teams of competent and friendly people. And Ania is a great PO.
Ania: Me and Kazik are not objective in the matter of Money Mapping because it is our project, in which we have been in from the very beginning, creating the concept, both from the technological and technical side. At first, there were just the two of us. Now our team consists of 13 people who work on various projects, currently in the area of the supply chain – also very interesting, because they quickly provide answers to real business needs.
Sandra: I have also congratulated already Kazik on his accepted proposal for PyCon PL ’22. Could you please tell more about your presentation?
Kazik: I talked about Variational Autoencoders. It is a model used to generate new data. For example it can generate new images of clothes – which do not exist in reality.
I found out PyCon PL ‘22 during researching what data science/Python conferences happen in Poland. Conferences are great opportunities to acquire knowledge of what people do both in research and industry. I believe it is quite important in the fast-evolving discipline of ML. What is more, conferences are a nice occasion to network with fellow data scientists.
Regarding PyCon, I have decided to take part as a speaker to gain new skills in technical presentation. I submitted a speech proposal and was accepted. Of course, it was a bit stressful to present for the first time, but form what I talked with conference participants the speech was interesting for them. Definitely doing speech rehearsal helped me to prepare. As far as I know, for the moment the presentation is available on the Internet but I suppose it should be available in the near future on PyCon PL Youtube channel since it was recorded.
Ania: Kazik is after the presentation, which was as usual very high level, his substantive knowledge combined with the willingness to share it is a real treasure for the team and the entire data science.
Sandra: I need to admit that Silky Coders is really doing great job being active online! When you are active on platforms such as LinkedIn it’s hard to miss all the content you’re constantly delivering about your workplace about the project or onboarding process. Also, it’s wonderful how your employees are your best promoters! Do you have any social marketing strategy tips you could share? What channels do you utilize?
Kazik: We have a great P&C team 🙂
Ania: We really like what we do and we respect each other which helps to create a nice workplace. In addition, the P&C team makes sure that we have good conditions, competitions and other attractions that build our team spirit. Once we won with Kazik as the fastest team to solve the puzzle in the escape room.
Sandra: Let’s touch quite controversial topic – AI taking over human roles fully or partially. How you feel about taking over the task of a stylist? Or at least supporting the design process? Has Silky coders already tried to automate the stylist logic? By this, I mean the thought process constantly assessed by the personal preferences of the client, taking into account their tastes, how they want to be perceived, and analyzing experiences with similar consumers. Great stylists bring both knowledge and the human factor to their work. They connect with their customers and help them find a fashion they will love while allowing them to go beyond their comfort zone. Do you think that with the right software, the world of tech in general is able to automate such skills and abilities which make stylists and designers essential in the fashion would?
Kazik: E-commerce recommendations that we do are kind of such a stylist. And we do it. But I’d say one thing is recommending someone items based on one’s preferences, completely other, and frankly much tougher is to create an ai tool for designers, able to support human designers. For the moment we haven’t done it. However looking at how the generative abilities of ai develop, I believe that we will have such tools supporting human designers in the future. What I think is definitely not possible is replacing human designers. Such a tool would just make their work easier.
Ania: We make recommendation systems, we are able to create systems supporting stylists, but I believe that an artistry and sense related to styling or art we will never be able to replace with artificial intelligence. Creativity and going beyond the limits is a great human strength.
Sandra: If there were no limits on the world, whole toolbox and all possible data available, what would be your dream project?
Kazik: Frankly, I quite enjoy what I am doing right now. I am doing my part to make things more optimized and less wasteful.
Ania: I like our projects, they are interesting and we see how much they help specific people in our business. What else would be a dream project for me? A project supporting the natural environment. I had the opportunity to collaborate on the Detect Waste project conducting by Women in Machine Learning and Data Science association. It would be nice to continue working on AI for good.
WIMLDS ‘s mission is to support and promote among others women interested in the fields of MLand DS. We have chapter here in Tricity (click here for more info) and I came across their initiative related to plastic detection and decided to use both my knowledge of the geotechnical industry, related to wastes, and knowledge in the field of data science.
Sandra: Any books/courses recommendation?
Kazik: Andrew Ng’s course on Deep Learning is definitely recommendable. Murphy’s Probabilistic Machine Learning, Aurelien Geron’s Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow and for the programming Lutz’s Python: Introduction. Oh, and maybe Machine Learning Design Patterns (Lakshmanan, Robinson, Munn). There are more but here are some I can definitely recommend.
Ania: I believe that it is worth being up-to-date, browsing websites related to data science and valuable content on Linkedin. If you are starting your adventure with data science, it is definitely easier to start with a specific course, e.g. at Data Camp. From books to start, Machine learning, Python and Data Science. Introduction, Andreas Muller. My part is Professional Product Owner, The: Leveraging Scrum as a Competitive Advantage, Don McGreal.
Sandra: What people do you seek for Silky Coders right now? What skills are most necessary?
Ania: Last year was very intense for us, we recruited over 200 people. We are currently expanding our e-commerce and mobile teams, and of course devops. The area of technology in which we operate is very wide, from Java, through PHP, to logistic systems, so we could list it for a long time. In addition to technical skills, at Silky Coders we are primarily looking for people who are focused on development.
I hope that you enjoyed our chat around data science for fashion brands as much as I did. If I had to pick just one thing that impressed me the most about my interlocutors it is definitely their passion. I truly believe that your job can be far more than the way of earning money and both Anna and Kazik with their stories, recommendations and after work activities are such great examples of this mindset. If you’re really passionate about something, you can be sure that sooner or later you’ll attract similar freaks (like Anna did with myself on InfoShare 🙂 ). That’s why I am such a big fan of communities and with all the good stuff shared with my old/new friends from Silky Coders I am ready to promote them further.
I am really curious how not only Silky Coders will scale in Poland but also how Anna’s and Kazik’s careers evolve! Keeping fingers crossed for both, looking forward next speaches and AI for good projects mentioned, I would like to thank both of them for their wonderful energy and ease of cooperation. And to you, my Dearest Readers I would like to ask about the topics you would like to read the next interviews on, and also invite to some previous conversations on the blog.
Sending warm wishes for a wonderful Christmas and a Happy New Year! See you in 2023. 🙂