My journey into data science – Alexey Grigorev, Principal Data Scientist, OLX Group

One just needs enough information to make a model work and not drown in every detail, thinking that at some point it will be useful.
Listen to this story

Alexey Grigorev is the Principal Data scientist at OLX group, a Dutch-domiciled online marketplace headquartered in Amsterdam. With a joint master’s degree in IT for business intelligence from Erasmus Mundus, Alexey specialised in ‘Distributed and Large-Scale Business Intelligence.’ Along with a stellar career in various firms like Auriga, Luxoft, Simplaex, etc., Alexey also founded DataTalksClub to provide a guiding platform for aspirants looking to make a career in data science.

In an exclusive interaction with Analytics India Magazine, he spoke about his data science journey and the challenges that he tackled in his vibrant career.

AIM: How did your data science journey begin? 

Alexey: Since my university days, I have always been fascinated by anything that involved data – statistics, time series analysis, etc. I also studied a bit of artificial intelligence in my college days and worked on projects involving clustering and time-series analytics, but that was university work. Due to a lack of opportunities in the data science field, I took up a visual developer job soon after graduation. I was working with Java, and eventually, I was introduced to the popular course by Andrew NG, which brought me back to the world of data science. This course gave me clarity that data science was indeed something I would like to take up seriously.

Luckily, by that time, there were multiple positions available in that space. I also pursued my masters in this field, although I’m not sure if it was the best time investment. I also worked on a few freelance projects in between. That’s how I started off in data science, and the journey has been fun these past seven years.

AIM: How important is it for aspirants to start early and develop the portfolio?

Alexey: Starting early and developing a portfolio, working on projects is of the utmost importance in this day and age. As I mentioned earlier, getting a master’s degree was not the best time investment because creating a portfolio and working on several projects would take a quarter of the time taken to get a degree. Granted, Universities provide a good environment to solidify one’s fundamental and networking opportunities. However, careerwise, investing two years would mean the difference between a junior data scientist and a mid-level data scientist or between a senior data scientist to a data science manager. 

AIM: What specific fields of your work do you find most fascinating?

Alexey: Ad-tech is a field that’s always fascinated me. Integrating ads, especially in between video/mobile games, is quite a challenging and interesting issue; if not dealt with properly, they can become annoying. It requires a ton of information and the capability to process all this information quickly. 

Another challenging aspect that piqued my interest is the efficiency of models post-deployment. AI/ML models need to be constantly updated, even post-deployment, as they continuously take in data and perform inferences on it. Maintaining efficiency and compute time is a challenging process, and one can learn many things about a model that was useful afterwards.

AIM: How did DataTalksClub happen?

Alexey: It was a natural process. Initially, people would reach out to me, seeking guidance related to career transitions and how to get into data science. I am quite active on LinkedIn and started giving career advice to a lot of aspirants in the form of hour-long sessions. In this process, I grew tired of answering the same questions over and over again. I could not ignore questions just because it has already been discussed, so I looked at how I could reach out to more people. 

That’s when the idea of creating a platform where all of these people came together in one place was born. So I created such a platform and scaled it to what is now DataTalksClub. This change came as I was striving to move from one-on-one conversations to one-to-many. That’s how the idea of Slack came to being. My vision was to go from one-to-many to many-to-many, where people whom I’ve helped would go on to help others, thus building a collaborative network. I met a lot of people I didn’t know before, and it’s been a fun journey.

AIM: How does OLX use AI/ML?

Alexey: OLX is a vast platform consisting of more than 40 ML services in production. For example, one of the core issues we address is moderation between sellers and buyers in the global marketplace. It is one of the first use cases of machine learning at OLX. Let’s say a seller wants to post the sale of an iPhone. He creates a listing but to get more viewers, and he posts multiple duplicate ads. OLX sifts through such duplicate ads and removes them in order to reduce spam. Another aspect of moderation is censoring explicit content, like if somebody is trying to sell a gun. From a buyer’s perspective, we at OLX strive to give people what they are looking for in the fastest way possible. In order to do that, we need a good recommender system that looks at your activity and interaction on the platform. Our search engines provide a personal touch based on interest and relevance of a search query to the actual item. 

Of course, I’m just scratching the surface, and there are a plethora of ML-based capabilities that we at OLX work with, making the overall experience better.

AIM: Your vision toward data science?

Alexey: My vision is that data science should be simple enough for everyone to use. Say a product manager wants to test an idea. In order to do that, the product manager will need a data scientist and a team of ML engineers who will create a working model. What I envision is that, ideally, the product manager should have the option just to open a platform, pull data from different sources and test his idea on his own. Custom models that are just a few clicks away. However, the product manager still needs to know some basic concepts of AI/ML like cross-validation, precision-recall, etc. However, they need not have technical expertise in areas such as Python coding, putting the code on Docker and deploying it on the cloud, etc. There is a lot of such no-code, low-code platforms that cater to such needs. The functionality is not up to the mark. Right now, we have a proof of concept. 

AIM: One piece of advice you would like to give to a data science aspirant?

Alexey: Learn by doing projects. It is better than learning from textbooks. Textbooks are important. After studying a textbook, one feels that they know everything. However, the application part is a whole different ballgame. In order to get a true sense of how things work, you need to build things. I’m a big proponent of building things and then understanding how it works. This approach works quite well for engineers as I am an engineer myself. Data science is a vast field, and mastering all of its fields is quite a daunting task. It is ok sometimes to not know everything. One needs enough information to make a model work and not drown in every detail, thinking that at some point, it will be useful.

Download our Mobile App

Kartik Wali
A writer by passion, Kartik strives to get a deep understanding of AI, Data analytics and its implementation on all walks of life. As a Senior Technology Journalist, Kartik looks forward to writing about the latest technological trends that transform the way of life!

Subscribe to our newsletter

Join our editors every weekday evening as they steer you through the most significant news of the day.
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.

Our Recent Stories

Our Upcoming Events

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox

6 IDEs Built for Rust

Rust IDEs aid efficient code development by offering features like code completion, syntax highlighting, linting, debugging tools, and code refactoring