Social Media is an interesting domain that entails a whole new dimension of challenge for Data Scientist. – Vaibhav Saxena

Interview with Vaibhav Saxena

Vaibhav Saxena

Data Scientist at Radius
Bangkok, Thailand

Q1. Please share your educational and professional journey?

I completed my school in 2012 at a very small private school called SBRL Academy in a very small town of Mainpuri, U.P. with some decent results. And I pursued my Bachelor of Technology the same year at Hindustan College of Science of Technology, Mathura, U.P. in the domain of Information Technology and completed it in 2016 with an on-campus job placement at Tata Consultancy Services.

I started my professional carer in 2016 with Tata Consultancy Services Ltd in 2016 as a Systems Engineer in 2016 in Chennai, TN and I served there for almost 3 years and switched to an SME in New Delhi called DMI Finance as a Data Scientist in 2019. DMI is primarily a FinTech company. After working for 6 months there, due to personal reasons, I moved to Bangkok after securing a job with a social media company called Radius in late 2019. I am working with Radius ever since.

Q2. What did attract you towards Data Science Domain?

Very briefly: NLP. In Chennai, I had a chance to attend a seminar by Virtusa in 2018 on how NLP is tailoring the user-experience online which showcased tasks the Language Modeling, Q&A and Tags Prediction which just completely caught me off-guard. I was working as a Java developer in IT support and I have never heard of solutions of that calibre can also be done by a non-researcher.

I started to dig deeper into Python and a moment of epiphany came when I made my first multi-class model using Random Forest Classifier on a dummy dataset to predict the income level of a person. At that moment, I knew I can’t not be working in this field. The passion grew on me over the time as I learned strategies to make a model understand the human language more and more precisely.

Q3. You are working in the Social Media Domain as a Data Scientist. How do you help as a Data Scientist in the Social Media Platforms?

Social Media is an interesting domain that entails a whole new dimension of challenge. The other domains do not have really force you to optimize your models in production to cut the inference time but in this very domain, this challenge in brutally pervasive. Making a good model (which is itself a challenge in Social Media) is one thing, making it respond lightening fast is totally another. As for the tasks, the data is full of user’s conversations in the form of posts, comments and messages.

You have to work on making the user’s experience better and come up with ingenious ideas on how to build on those ideas and complete the tasks with minimal cost incurred. Tasks like delivering the perfect search results, recommending the most related friends/pages to users, text autofill, control hatespeech are notable mentions. However, the key is the optimization, not only how good your model is but also how fast is it.

Q4. How do you motivate yourself at every morning?

Working in a foreign country in a very busy social media company allows me to stay swamped most of the time. The projects are challenging which demand a fair bit of research especially with state-of-the-art algorithms. Luckily, time is not a restraint with me. Every morning, I recall my last day’s work and figure out some solutions in my head that I plan to implement that day and progress with the task.

Q5. What do you think about Online Data Science Courses? Do you update yourself in terms of education?

The online courses are the ones that made me learn and crack an interview. They are incredibly helpful by any means. In the beginning I was solely relying on online courses ranging from Udemy (Jose Portilla, Lazy Programmer, Kirill Ereminko), Coursera (Stanford) and YouTube for many free videos. After 3 years, I have delved more into the construction side of the models focusing more on Maths, research papers and the updates in the DS community on Reddit, Kaggle, Twitter and LinkedIn.

The new NLP courses under NLP Specialization on Coursera are extremely good to hone in on some important basics. I am also finding all the opportunities to learn state-of-the-art algorithms online like BERT, T5, DETR, EfficientDET etc. My go-to form of learning today is still online Data Science courses.

Q6. Where do you want to see yourself after 10 years in career?

I have a passion for finance especially wealth management and I learn finance all the time and ways to grow money.

I am getting better day-by-day into both domains and I wish to start a business built around wealth management and backboned by AI.

Q7. Which one thing do you want to change in yourself and why?

I have never been a team lead only until recently I have to lead a team of 2 really well equipped Data Scientists who completed their Masters in Applied Sciences from Melbourne University. This was at first a challenge to manage but I have been learning ever since.

One thing I have to develop the most is to have the ability to discuss the activities in a manner of constructive criticism especially to my peers. I also have to learn on how to get the best out of my team by also keeping the professional relationship with them as healthy as possible.

Q8. Which is your most used Machine Learning Algorithm/Model in your career to solve real world problems?

The text data that I deal with is majorly in Thai. I have developed an effective CPU-powered CNN model to provide text suggestions while searching for anything on the application. It responses in 2ms and delivers the most related best matches in the suggestions while typing.

This one I am most proud of because of the efficiency of the model and also the fact that the model understands Thai which in itself a big task given the language is one of the most uniquely expressed languages I have ever seen.

Q9. Which one skill do you like most about yourself?

I like negative feedbacks on my work more than the positive ones. My supervisor knows this and he makes sure he is ready with some shortcomings with constructive criticism.

Get more about Vaibhav Saxena:


Data Scientist Interviews : Abhishek Thakur Interview, Krish Naik Interview and many more.

Leave a Reply

Your email address will not be published. Required fields are marked *