In the Author Spotlight series, TDS Editors chat with members of our community about their career path in data science and AI, their writing, and their sources of inspiration. Today, we’re thrilled to share our conversation with Vyacheslav Efimov.
Vyacheslav is a Senior Machine Learning Engineer specialising in NLP and Computer Vision. One of his passions is making educational content: Vyacheslav has published over 60 articles on Towards Data Science, explaining complex concepts in simple words, providing rich visualizations.
You wrote many beginner and explanatory articles on TDS. Has teaching the fundamentals changed how you design or debug real systems at work?
I notice the correlation between the more I teach something, the better I understand it. In real life, when I write a new article, I strive to dive into small details while keeping the explanation simple for my readers. Emphasizing information in this way helps me better understand the workflow of algorithms.
In that sense, whenever I encounter an error in one of the algorithms used at work on which I wrote an article in the past, there is a higher chance that I will rapidly figure out the solution to the problem by myself. From another perspective, when I write an article on an unfamiliar topic and explore it myself, it increases my confidence when I apply that particular algorithm at work, as I already know its application scope, advantages, disadvantages, and specific details or constraints.
This way, I can come up with original solutions that are not obvious to others and back up my choice to other teammates, managers, or stakeholders. That knowledge is precious to me.
With so many new models popping up every day, it’s easy to feel completely swamped. How do you decide what’s worth a ‘deep dive’ and what you just ‘get the gist of’? Has your strategy for managing this changed at all recently?
Today, we indeed have an abundance of models and tools that appear every day. It is easy to feel lost when you’re unsure about what to pursue next.
With limited time, I usually delve deeper into topics that may be applicable at work or in my personal projects. This gives me more confidence when I have to present or explain my results.
Businesses usually want to achieve working results as soon as possible. This is also one of the reasons why, in my articles, I focus more on theoretical concepts, as I cannot devote my time at work to going into theoretical depth.
This way, I have an efficient blend of practical experience at work and theoretical insights in my blog. Both of these components are important for skilled data scientists.
You’ve competed in AI hackathons. What did you learn from having such tight deadlines? Did it force you to get better at scoping projects or deciding on a model? And do you find yourself using any of those ‘hackathon lessons’ when you’re sketching out a new idea from scratch?
Hackathons typically last between several hours and two days. That is a very small period of time to develop a fully functional product. However, at the same time, it pushed me a lot in the past to better prioritize the features on which I should focus. In general, time management is a valuable skill to have. When you have several possible solutions to address your problem, you must choose the one that best fits the business needs while also respecting time constraints.
What is also great is that after every hackathon, you can evaluate yourself in terms of the time it took you to implement certain features. For example, let’s say that it was the first time you had to develop a RAG pipeline, which took you around 4 hours to implement. The next time you face an analogous problem at work or a hackathon, you will have a better estimate in advance of how much time it would take if you decide to use the same method. In that sense, the hackathon experience allows you to better define time limits for the methods you want to implement in projects.
For me, the biggest lesson from the hackathon was not focusing on perfection when creating the MVP. While an MVP is important, it is also necessary to present your product attractively to clients or investors, explain its business value, the problem it solves, and why it is better than existing solutions on the market. In this regard, hackathons teach you to come up with better ideas that solve real problems while also shipping the MVP quickly, containing the most essential features.
For readers thinking about their career path: your “Roadmap to Becoming a Data Scientist” series spans fundamentals through advanced ML. If you were rewriting it today, what topics would get promoted, demoted, or cut entirely, and why?
I wrote this article series a year ago. For me, all the concepts and topics I listed are up to date for aspiring data scientists. All math, computer science, and machine learning topics I present there are an essential foundation for any machine learning engineer.
As we’re now in late 2025, I would also add a requirement to have at least minimal experience with prompt engineering and to be familiar with some AI-generative tools, such as GitHub Copilot, Gemini CLI, and Cursor, which would allow for increased work efficiency.
As a note, compared to previous years, IT companies have higher requirements and expectations for junior engineers entering the data science field. It makes sense, as modern AI tools can perform junior-level tasks very well, and many companies prefer to rely on them now rather than on entry-level engineers, as they don’t have to pay salaries while in both cases they receive the same result.
That is why, if a machine learning engineer possesses the strong fundamental skills I described in that series of articles, it will be much easier for them to dive autonomously into more complex topics.
Your background blends software engineering and ML. How does that foundation shape the way you write?
Having strong software engineering skills is one of the best advantages you can have as a Data Scientist:
- It makes you realize the importance of well-structured software documentation and creating reproducible ML pipelines.
- You understand better how to make your code clean and readable for others.
- You understand algorithmic constraints and which data structure to choose for a particular task, based on system needs.
- You can more easily collaborate with backend and DevOps engineers on integrating your code modules.
- You do not need to rely on others to make SQL queries to retrieve necessary data from the database.
The list can go on and on…
Speaking of my articles, I don’t have many that present a lot of code. However, whenever I do, I strive to make it readable and understandable to others. I always put myself in the shoes of others and ask myself how my article text or code examples would be easy to perceive or reproduce if I were in others’ shoes. This is where the software engineering experience makes this realization more essential for me, and I follow the best established practices to deliver my final product.
Looking at your portfolio and GitHub, you’ve blended software engineering fundamentals with ML from the start. What’s one engineering habit you wish more aspiring data scientists adopted early?
Many engineers, especially juniors, tend to underestimate the importance of creating good documentation and reproducible pipelines. This also happened to me in the past, when I was more focused on developing robust models or conducting research.
As it turned out, when I had to change contexts and then several weeks later to return to work on the previous project, I was then spending a lot of time figuring out how to run my old code in a messy Jupyter Notebook or install necessary libraries again, where I could have just spent a little more time in the past by developing a well-documented README.md explaining all the required steps to execute pipelines from zero.
Because it was nearly impossible to rerun my pipelines from scratch, I was also unable to conduct experiments using other entry parameters, which made the situation even more frustrating.
It was a painful experience for me, yet one of the most valuable lessons I’ve learned. So if I had to give a piece of advice to an aspiring data scientist on one particular habit, it would be this:
“Always make your machine learning pipelines reusable and well-documented”.
Over the past year, has AI meaningfully changed how you work day to day as an ML Engineer? What got easier, what got harder, and what stayed the same?
In recent years, we have observed a significant rise in powerful AI engineering tools:
- LLMs, which can respond to almost any question, give advice, or find bugs in software
- Cursor, Lovable, and Bolt are acting as AI-powered IDEs for developers
- AI agents can complete multi-step tasks
As a machine learning engineer, it is essential for me to regularly adapt to these tools to use them efficiently.
What became easier
Starting from 2025, I can observe the following positive impact on my work:
- For me, it became easier to rapidly test ideas or prototypes. For example, there were times at work when I was given computer vision problems that fell outside my area of knowledge. In this way, I could ask ChatGPT to propose several ideas to solve them. There were even times when ChatGPT generated code for me, and I tried to execute it without understanding how it worked inside.
Then I had two possible cases:- If the code ran successfully and solved the initial problem, then I tried to go deeper inside the OpenCV documentation to understand how it ultimately works.
- If the code didn’t solve my problem, I would either ignore it, report the error to ChatGPT, or attempt to find the solution myself.
As you can see, I was able to rapidly test a solution that could work and save me hours of research without any risk.
- Another excellent use case for me was inserting error messages directly into ChatGPT instead of searching for a solution on the Internet. It worked well most of the time, but sometimes it was affected by errors related to library installations, system errors, and the deployment of pipelines on the Cloud, among other issues.
- Finally, I am a big fan of AI hackathons! Having tools that can generate both the frontend and backend of your system makes a massive difference for me, as I can now rapidly create prototypes and test my MVP in several hours. What I develop now during one-day hackathons could require an entire week of work.
What became harder / risky
- When writing code with AI, there is a higher possibility of sensitive data leaks. Imagine you have a file or code fragment containing essential credentials that you accidentally feed into an AI model. Then a third-party tool will know your sensitive credentials. It can happen, especially if you use a tool like Cursor and store your credentials in another file rather than .env. As a consequence, it is always necessary to be very cautious.
- Another risk is not properly testing the AI-generated code and not knowing how to make a rollback. An AI tool can introduce invisible errors in the code, particularly when it is used to modify or refactor existing code. To ensure that AI-generated code does not degrade, it is necessary to thoroughly review the generated code parts, test them, and save modifications in a way that allows you to always rollback to a previous, correct version if necessary.
- When relying too heavily on generative AI tools, there is a risk that the code will become unreadable, contain excessively long functions, exhibit repetition, or cease to function correctly. That is why it is essential to understand that AI tools work more effectively on prototyping than on maintaining high-quality production code.
What remained the same
What remains constant for me is the importance of understanding the internal workflow of the algorithms I use, maintaining strong computer science foundations, and writing high-quality code, among other key skills. In other words, the basic principles of software development will always be necessary to efficiently use AI tools.
In that sense, I like comparing a set of available AI tools to a substitute for a junior developer in my team, to whom I can delegate less essential tasks. I can ask it whatever I want, but I cannot be 100% sure it will do my tasks correctly, and this is where the importance of having strong fundamental expertise comes into play.
To learn more about Vyacheslav‘s work and stay up-to-date with his latest articles, you can follow him on TDS or LinkedIn.
