Your DS Story S01E10: Saurabh Goel (Prudential Financial)
‘Your DS Story’ is my attempt to bridge the gap between data science professionals & data science aspirants. Here new crop of data scientists will share their experiences, struggles, achievements & their advice so that data science aspirants/enthusiasts can learn and get inspired.
Saurabh Goel is an experienced senior data scientist who loves working with data and try to help businesses improve their performance along the way.
1. Please tell us a bit about your background?
First of all, I would like to thank Ankit Rathi for taking this great initiative and giving me this opportunity to share my experience with the entire DS community. Hopefully I would be able to inspire some budding data scientists who are trying hard to come into this space.
I am a typical product of Indian education system with BE (Electrical, BVP Pune) + MBA (Finance, IBS Hyderabad) under my armour, started my career with Cognizant Technology Solutions (CTS) post engineering. Worked with CTS for 2 years before pursuing my MBA with IBS. I got exposed to analytics post-MBA and currently, I am working as a senior data scientist with Prudential Financial, Letterkenny (Ireland). In my current role, we help various business function across Prudential to solve their problems with the use of data. Recently, I have also started my MSc in Artificial Intelligence from the University Of Limerick, Ireland as well.
Everything about data science excites me:
Solving actual business problems and see their dollar impact
Ability to see the future with good accuracy
Steep learning curve due to ever-changing technologies
Presenting the work to business leaders across organizations
Seeing the models built put into production
2. What projects you are working these days?
I have worked across various data science problems, however, my main interest and expertise lies in Insurance domain.
Some of the recent projects I have worked on are:
Developing risk segmentation strategies for a P&C insurance carrier to underwrite the better risk
Finding causes of water damage claims in the US from claim notes using NLP algorithms
Developing market strategies for leading telecom provider in the US
3. How your day to day job looks like?
For DS profiles, day to day job is actually very different from what is advertised online. The fancy part of the model building comes very late in the project and generally, that is not the toughest part of the job. In my understanding, the most difficult part is to formalize the entire business problem into analytics problem and lay out step by step what all needs to done to solve that.
After that, the next step is to collect and bring data into a shape which can be analyzed. There is no good data in this world, its either bad data or very bad data. The most important job for a data scientist is to own this bad data and convert it into something which can be analyzed keeping the end objective in mind.
I typically spend 70–80% of my time in data collection, data wrangling, data analysis and insights generation etc. All business stakeholders always want to see what is happening in their current portfolio before trying to predict the future using modelling techniques.
4. How you started with DS or transitioned into DS?
I used to love finance while doing my MBA and wanted a job in that space only. Fortunately, in 2011, I could not land a finance job and started as a business analyst in an IT firm based on my previous experience. Here, I got exposed to insurance domain which immediately attracted me due to such deep domain knowledge to be acquired. However, at the back end my quench for a number-crunching job was not leaving me, so I tried to find if I can do a similar thing in Insurance and voila!! I found actuarial science there. Actuarial science was the perfect answer for me at that time.
I started studying for actuarial exams while doing that research I got to know about data analytics as an upcoming field. This is 2011, where we did not have so much knowledge as we have today, but still, I felt overwhelmed by the amount of hype around analytics that was there on the internet at that time. So it was data analytics and actuarial in parallel happening for me as both could have given a job that I was aspiring for.
At that time there were very limited companies doing analytics, the primary one being Fractal Analytics/ MuSigma/ Genpact/ XL Catlin etc. I tried applying to all of these but all in vain as I did not have IIT degrees or Statistics degree. But I never gave up. I kept on saying to myself that I have to do analytics. Luckily I got to know about a walk-in drive at Fractal Analytics through some friend, Fractal had bagged a big insurance logo and they were looking for someone with insurance domain understanding. That day was one of the best days of my life when I got a job at Fractal based on my domain knowledge even though I did not have an iota of analytics knowledge.
After that it was a roller coaster ride, I started learning and reading day and night, learnt from each and everyone around me. I would always thank Fractal Analytics for developing a data scientist from a person who was not even a data analyst.
5. What advice would you like to give to DS starters or DS transitioners?
Few points that I can think of are:
Be a problem solver first, you should be able to see the end goal of the problem rather than thinking shallow in terms of modelling etc. Once you are able to see the end goal, path to end goal will start to emerge.
Get your basics right — if you are starting to learn something new like regression, read it from various sources — listen from various sources — understand the mathematical intuition. Without proper time investment do not expect to get the basic right but once you have them they will stick with you always. I have slept countless nights to understand basic algorithms from scratch but that is always a one-time effort, so put that effort.
Be a lifelong learner, no matter what, your learning should always be on the top priority simply because of the nature of the job.
Pick any language and try to master it.
Try to get a data analyst job at first, once you start working with actual data then you will be able to appreciate the data science pipeline much better.
Pick a mentor, everyone needs help. Your DS journey can be hugely shortened with the help of a mentor.
Data science is always used to help the business; it can never be the other way around. So any perfect model is useless for business if that is not generating dollars. Hence, always think in terms of business benefit rather in terms of algorithms.
Thank you for reading my post. I regularly write about Data & Technology on LinkedIn & Medium. If you would like to read my future posts then simply ‘Connect’ or ‘Follow’. Also feel free to listen to me on SoundCloud.