Data Scientist at Miro. (previously – Information Architect)

Moscow, Russia. Amsterdam, Netherlands.

Online collaboration platform (aka AI-powered digital whiteboard).

I develop a special kind of logic (called models) which, when incorporated into the digital tool, helps users to:

• either automate their routine tasks (e.g. cluster some collected feedback by topic);
• or to personalize their workspace by proposing tools and actions that are relevant specifically to them at this particular point of the time (e.g. suggest a relevant template to start their work with).

• Shape the goal, product, and technical limitations as well as metrics with the team
• Collect data that is relevant to the task (without breaking privacy policies!) and clean it up from duplicates and inconsistencies
• Transform this data to be in a form that makes sense for models (basically, into some kind of numerical format)
• Train different kinds of models on it to select the best-performing
• Wrap up the best-performing model into a logic for it to be ready to work in a production environment

Hard skills:

• Bunch of coding skills: python, spark, sql at least
• Machine learning skills: ways to get and preprocess the data, different kinds of models, and ways to optimize their training
• Some (the more the better) knowledge of infrastructure and software engineering: how to set up relevant data collection, how to store and experiment with it safely, ways to deploy models to production, track their performance, and schedule retrainings.

Soft skills:

• Communication
• Collaboration
• Iterations

Small pieces of data, not very important and informative by themselves, can be aggregated and used to build some really valuable functionality, helping this particular person.

The process of getting proper data is a nightmare. But actually, I love it.

Training models – is a cherry on the top of the cake. Data collection and pre-processing is the cake itself. With a quite sophisticated recipe.

Deep knowledge and (regular) hands-on experience in Data Science related technical domains – Data Engineering and Machine Learning Engineering.

I am acquiring Data Engineering skills.

Complicated, pretending to be close-to-real-world approaches rarely work better, than simple ones. If they do, the difference is not dramatic and isn't worth the time spent.

As a former Information Architect, I am very attentive to the data – I don't hesitate to put effort into making it consistent and representative for the business case aimed to be solved.

Having a background in product, I work taking into account product features development workflows as well as future user experience, and use this knowledge for better prioritization and neater cross-functional communication

Well. If I compare my current self with my future-wannabe self, I would give myself a 6. Probably the score on an industry scale would be a bit better.

Technical skills related to Data Engineering (building data lakes, data warehouses, pipelines, event processing systems) and ML-Engineering (model deployment options, model acceleration tricks, touching guts of existing models and ML-relevant libs, etc).

The extent at which I enjoy the domain I am working in :)

• Databricks/Jupiter
• Snowflake
• Airflow
• Miro
• VSCode :)

I try to group meetings as much as possible, to avoid letting them split my focus time. I also use placeholders for focus time in the calendar.

This highly depends on the mindset, but I personally prefer not to switch topics/projects during the day, and in general, also try to have no more than two projects running in parallel.

Tending to be detail-oriented, I try to put myself to a meta-level once a week and to reflect on whether I am going in the right direction, taking into account goals and timelines.

Well, it is definitely important, but being an introvert, I prefer rare, but deep conversations on subjects that are most relevant to me or to my counterpart at this point, over shallow, broad, and frequent.

Since writing down in a structured way something I learned firms my understanding, I tend to write and share blog posts with summaries or project details from time to time. This sometimes brings relevant people and generates meaningful discussions

In such a fast-evolving field as Data, it is also important to keep track of things going on. But I don't want to be "stack overflowed" either, so I try to carefully select people from the industry I follow, whose curation principles and stylistics resonate with me.

Information hygiene is the boss :)

I am lucky to have the domain I am working in to be my hobby and passion at the same time. So I would say I have a work / out-of-work work balance which is sometimes interrupted by other unavoidable matters.

The same, but probably with the possibility to spend more time on the acquisition of more hard-core technical skills and knowledge.

And woodworking and welding – for occasional digital detoxication ;)

The answer is "it depends". For myself I don't need that much on top of my paid bills and supermarket basket, but then people you are accountable for come in, and the situation changes.

Can't remember anything specific.

Learn quicker how does specifically your brain work, to optimize your educational and professional trajectory.

Start proper CS education right after you wrap up with your Biochemistry, this is your stuff!

I wouldn't give any advice if I wouldn't be asked for it ;)

But if I would, before giving any advice I would ask why they think they want this work. And what exactly do they think they would enjoy? Depending on the answers advice would be different.

I am not the kind of person to be easily excited by trends. I prefer to get excited by things that remain in place and become a building block of the domain after the trend changes.

• Chat GPT as a great counterpart to bother with stupid professional questions (critically considering the responses tho);
• Braun items by Dieter Rams;
• Muji items;
• Makita power tools;
• Volvo 740 wagon

• The Information: A History, a Theory, a Flood – book by James Gleick
• 3Blue1Brown – youtube channel covering different math topics