Data science is the most talked about profession today, and rightly so!
Its immense potential to unlock the abilities of data to augment business decisions and create intelligent digital products is capable of replacing human effort, making life better and easier!
But with much chatter comes a clutter of information that can confuse anyone and everyone. If you are an aspiring Data Scientist struggling to find a way, here’s something for you!
To bring authentic expert advice to you, we are in conversation with Narayan Sharma, Data Scientist at Naukri.com. Read on as he shares his experiences, insights, and advice from his career journey!
Could you share a little about what you do?
I work as a Data Scientist with the Naukri Recruiter (Resdex) team. So, Naukri has two primary sources of revenue: The Naukri Recruiter and the Jobseeker, where candidates register to apply for job postings made by these recruiters.
A little background about Resdex- it is a recruiter-centric business with a database of 10Cr+ jobseekers that powers the recruiter search.
However, with an ever-increasing user base, it becomes incrementally hard to find the best matches for recruiters quickly.
Now, here is where I come in – Building Recommendations! You would have seen a little section "People who bought this also bought this" or "See more items like these" on e-commerce platforms- those are what we call recommendations.
I build similar candidate recommendations for recruiters to discover and shop job seekers more quickly and easily.
Further, these recommendations can create personalization modules or even (re)calibrate the search process. I do a bit of all-building, maintaining, and owning “All things Recommendations” for Naukri Recruiters.
Simply put, I ease the candidate discovery process for recruiters. For example, if a recruiter wanted to find a software engineer with 8+ years of experience based out of Delhi, they would get 3L+ results.
Now, if they find and like one user profile, they can click on its recommendations and find 100 other people with similar profiles. So, that is what I build.
How long have you been working as a Data Scientist? What piqued your interest in the field?
It has been almost two years with Naukri. Previously, I worked as a Data Engineer turned Software Developer for a year and a half. So, it has been around 3+ years with data and software products.
About what piqued my interest? That is an interesting story!
I realized early on that I won't make much of a living by selling stuff made by someone else. Also, I could never sit in front of a screen and call myself a hacker. So, the obvious choice was to find a middle ground.
I liked computing but not core software development, enjoyed statistics but not math. I loved storytelling but wasn't much of a reader and liked graphs but skipped the theory that explained them.
Data Science happened as some sort of Ikigai (Japanese for life's purpose).
During the second year of college, I discovered that some people were trying to do visualizations based on data. That was my inlet into what they now call Data Science.
I was aiming to be a Business Analyst, not a Data Scientist; I just wanted to use R, Excel, or Tableau to make those graphs. But after doing a bit of visualization, creating some beautiful charts, and sending them across, I realized it was not working out for me professionally.
One thing led to another, and a few courses later, I was introduced to Machine Learning (an extension of software engineering). After this discovery, I went ahead and took up some online courses.
By the time I finished my Fourth year, I had landed my first job in Data Science at a digital marketing firm. However, I had to quit as I realized it was more of a Business Analyst role instead of a Data Scientist.
I realized that the only way of getting into core Data Science is by first understanding the core of Software Development. That is when I took a detour, got into Software Engineering, and after a year of work, decided on my Master's degree.
Finally, Machine Learning happened.
How did you land your first break?
In my second year of college, I began fancying the idea of data visualization and calling myself a Data Scientist. Much like other freshers, I struggled to get recognized despite many efforts.
I am from Lucknow and was there when I heard that HCL was setting up a big Tech Park in the city. I reached out to a few people there and finally heard back.
The official informed me that it was more of a call center, and the only data available was caller data, and they neither had any data science nor data analytics team in place.
I said, “No problem! I will do something with your data, just let me see it!”
This was the breakthrough for me, and I wanted to make the most of it.
So, they set aside some CSV dumps, and I started my work. I remember taking my online classes in the morning and implementing the learnings at work in the evening.
It was a 40 people team at the time, and I assessed their calling times, availability, and performance accordingly.
I first created a benchmark index informing them of the optimum amount of calls a person should attend before they can finally call it a day.
Later, I built a small model that could predict if the caller would convert based on call data and attributes.
I went back to college, did some more reading, and charted a more directional approach, as explained earlier.
What are the three things you learned since working as a Data Scientist?
Do not fear the data or fret about it. It may look intimidating, but you will always get around it. The data will tell you what needs to be done.
You are only as good as your team! Without mentors, I would be nothing, no matter the online courses, blogs, or books. Nothing replaces hands-on expert guidance.
Understanding the domain and the business context of a problem is more vital than knowing how to code.
Coding comes easy, many people have already tried implementing what you want, and in all likelihood, I would always figure out a way to code it. But context and reason helps you identify how and what exactly to code.
What are the three things that you like & dislike about being a Data Scientist?
What I like the most about being a Data Scientist is that I contribute significantly to the core business of my company.
I know that what I'm building is there to augment critical functions of the business.
We are trying to optimize the approach of recruiter searches for candidates on Resdex, which in turn, gets you more customer satisfaction and adds directly to revenue generation.
The second thing is that the field is very demanding and constantly evolving.
So, if you can stick your head in and stay disciplined, there is no doubt that you will grow in your career and profession because there is nothing else in the world that will not dance to the tunes of Data.
The third thing I like the most about data science is that the gratification cycle (for me) is fairly small.
If you can build a small utility quickly and successfully, you will get noticed, but will also get twice as much attention for failing.
Things move fast, break, get fixed, break again and get fixed again and start showing impact almost immediately. It is a high-speed, high-intensity, and high-visibility field.
Speaking of dislikes, the first one is cleaning data. There is no standard practice of maintaining data for anyone in the world, and 99% of our job is to clean data, which takes up a lot of our time.
It is one of the most crucial parts of our job that people barely know.
Secondly, I don't like how people perceive this field and my job. It is challenging to make a person understand what I do, and it is not as cool or easy as people perceive it.
A lot of hard work, constant effort, and labor goes into this field.
These clickbait advertisements of online courses claiming to train a person to become a Data Scientist in 3 or 6 months diminish the entire idea of chasing perfection with Data.
So, I do not appreciate this connotation attached to how you can pivot into data science.
Lastly, I would say too many meetings, given the wide range of stakeholders at play. I feel this takes away the focus from the tasks at hand sometimes.
Despite that, it exposes you to several dimensions of the business.
How do you maintain a work-life balance?
The thing with work-life balance is that it depends on the kind of place you are in, not the vertical (DS/Product/Technology, etc.) you work within.
Data science is not a taxing field, but if your firm is chasing a challenging target, then yes! You will also feel the brunt of it.
For example, the primary difference between a young startup and an established firm regarding work-life balance is in its ‘culture of data.’
Data Science as a field thrives in the presence of three factors: ample amount of data, sufficient expertise to research and implement techniques written in pre-calculus math, and the culture of repeated iterations, trying and failing and trying again.
These factors are present in established firms like Naukri, and I believe Data science is not a field for individual contributors because every person perceives data differently.
You are only as good as your team. The major reason for Naukri to be able to provide a good work-life balance is that we have an amazing team with experienced people who have your back at all times.
So yes, work is rarely delayed, and I get to use the rest of my day to ensure I get my share of proper sleep, workout, and family time.
Could you share one least spoken fact about the field of Data Science?
One of the least spoken facts is that you need a lot of fundamental programming in Data Science.
To begin with, I think Data Science is a misnomer because I feel somewhere, people have blurred the margins around Statistical analysis, Business Analysis, Data Science, and Machine Learning these days.
If you look at a global Tech firm like the MAANG, you will find that their Job description for a Data Scientist is similar to a more advanced form of Business analyst.
Now, I think our perception of Data Scientists is different from that of their western counterparts. In some of these firms, we have MLE, which uses Machine Learning for intelligent computing as a resource to augment their software engineering practices.
Both start with Data and head the same way until a T-point, where one branches to the Left, which is about service-based analytical, statistical modeling, business analysis, and data science.
The other branches to the Right, which is more software-centric, like computing, building intelligent product features, Machine Learning, deep learning, computer vision, and natural language processing.
Now answering the question, the least said fact about the right side of this game is that you need to have a fundamental grasp of engineering because apart from working with data, you have to work a lot on your fundamental programming logic too.
On the other hand, I think the left side of the business demands more statistical rigor, creativity and attention to detail; the presentation skills matter much more and one needs to be forever-ready to hop in and out of domains.
The least spoken fact about this side is that its really hard to stand out here, with just one's technical skills. One needs to be a wholesome data storyteller along with having sufficiently high proficiency in dealing with numbers.
To believe that it is easy to understand the intricacies of these concepts would be the biggest flaw in judgment, simply because the stack of things to know is gigantic, and it is impossible to do so.
So, people must focus on knowing what they want to do and what space they want to end up in because the margins are blurred all around.
How do you differentiate between Business Analytics and Data Science?
After years of answering this question to myself and people alike, I finally figured out the right way.
In my opinion, if you use the data to capture trends, identify insights from a large pool of information, or use fundamental statistical techniques to augment decision-making for business, then that's Analytics.
But the moment you use that same set of data to try and build computational logic, which can be used as a feature in any mobile/web application or just some kind of a software product, then that becomes Machine Learning or more colloquially Data Science, and that is the broad divide.
I think it is engineering the data versus wrangling it that makes the difference. If you dive deep into it, many people give insights, work with those who work on those insights, create features, and vice versa.
So, it is safe to say, analysts and Data Scientists work hand-in-hand.
To sum it up, utilizing data for descriptive factors is Analytics, and utilizing data for predictive and prescriptive purposes transforms it into modern-day Data Science or Machine Learning.
How important is upskilling in the field of Data science?
Many underestimate the amount of perpetual learning the field demands. Most upskilling courses online for digital marketing or even software development teach you the fundamentals.
Except for a few changes in terms of tools, the logic remains the same. Mostly, your experience will help you look at the larger perspective while doing the same tasks.
For example, there are two Developers- one with 3 years of experience and the other with 4, and both have to build an API.
Now, the API built by the Developer with 3 years of work experience will lack the finer nuances that a 4-year experienced developer can bring to the table.
However, the field of Data science is changing so rapidly that the current models or tools could be rendered outdated in a span as short as the next 6 months. This is why Data scientists must be super proactive in upskilling themselves.
What would your advice be for aspiring data science professionals planning any particular core skills?
Before anything else, you must decide your future course-is it advanced analytics or engineering-centric computing?
If it's the latter, here are the four things you need to undertake:
- Work on your theoretical knowledge of statistics
- Fundamental Python programming (CodingNinjas, HackerRank, Medium Leetcode); Knowing a Query Language is always a plus!
- Do tons of Applied Machine Learning problems (Kaggle, Driven Data, Analytics Vidhya)
- Read fundamental theoretical Deep Learning and build beginner projects around it
Become well-versed in your fundamentals for all four, as you will need to use their combinations and work with these daily.
Also, please steer clear of the delusion that shortcuts can help you become a Data Scientist. There is no easy way to succeed. So, be persistent and disciplined.
Take small steps, give yourself a timeline, a minimum of six to nine months to study dedicatedly, and structure the basics thoroughly.
Lots of fundamental, object oriented programming, applied classical machine learning, theoretical and practical deep learning and crystal clear understanding of applied statistics - that's your recipe for success. There are resources galore on the internet for the same.
Find yourself a good mentor instead of stumbling across blindly. Industry experts or academy experts can guide you in the right direction as they speak from their expertise.
All you have to do is reach out!
Send a request and draft a simple message, and they will surely get back to you. You can also check out websites like Scaler where you can find many experts who can mentor for free and guide you equally well.
How has Naukri helped in your career growth?
Naukri gave me wings!
As a data professional, the biggest problem that I faced has been quite a bit of imposter syndrome. But, for a person like myself to have the opportunity to work with such an amazing team is a great start!
Naukri helped me overcome my insecurities by providing a team culture where it was easy to approach anyone with queries or suggestions, which encouraged me to grow.
Also, we were given meaningful projects almost immediately after joining, which helped us realize that our efforts at work have a visible impact on the business, thus raising the stakes and bringing about a sense of responsibility almost immediately.
So, the stakes at which I am working here and the trust that management shows in us are immense.
The third thing that Naukri gave me is recognition. The kind of people reaching out to me today as aspirants willing to pursue this field has skyrocketed, and I think the biggest problem that any data science aspirant faces are that at the lowermost levels, there is too much supply with very little demand.
Having made it into the industry with a brand like Naukri and then working on a product that affects almost 8 crore people every day gives me a sense of the impact that most of my counterparts across other firms don’t find too easy to emulate.
What is "In Conversation with"?
Seeking an expert's opinion is not always easy, so we are bringing the expert's opinion to you!
In this series, we aim to explore the career journey of people from different fields, backgrounds, and career stages who have tread the path and made a place for themselves in their specific areas.
Join us as we try to understand all about what they do, how they started, and where they are headed next.
Stay tuned to read more such articles. To get insights into your desired field, send suggestions to firstname.lastname@example.org!