How to be a data scientist, according to Grab’s data science head

“Focus on fundamentals like statistics. Have confidence in how you understand the data and what it’s trying to tell you.”

What do data science, data analytics, and business intelligence mean at Grab and how are they being used? — Wong Mun

The data science team takes care of the science. We build algorithms and models, and in general, translate research (existing and new) into applicable product features. So, from the moment a passenger opens the Grab app to the time a vehicle arrives, data science powers the thinking and decision-making on the most efficient routes, travel time, and price points.

Data analytics looks at data from multiple sources to find trends and patterns—these insights can be turned into business decisions. Business intelligence looks a lot at our internal operational data to find ways we can improve our business processes, operations, and decision-making.

What does it take to be a data engineer/scientist? — Fu Chen Chia

Data engineers need to take care of data warehousing, building pipelines, and ensuring availability. Other than being familiar with the current state of the art, they must also constantly be thinking about adopting newer big data technologies to help them continue to scale.

Data scientists, on the other hand, need to find problems in the business, ask questions about the problems, find data, build models/algorithms to solve the problems, and validate the solutions. There are varying levels of science that you could use to derive different levels of quality in your solutions and results.

I am in my early 40s and just discovered ML in Apple WWDC last year. I was mind-blown. Any advice on how to break into the ML/DL career path? I’m a firmware engineer by training. — Terrence Goh

Career switches are always difficult. Fortunately, firmware and algorithms aren’t too far apart. Consider building intelligent embedded systems? I think that would be a good and natural next step. And since you work on firmware, it also shouldn’t be too difficult for you to understand how GPUs work or process data. Try to connect the dots, so you can move from one to the other.

How did you get into this career path? — Terrence Goh

I was fortunate to have training in science and engineering, then a stint in a startup that had to deal with lots of data, at a time when big data was getting more and more recognition. So, I was fortunate, and I’m grateful. 🙂

As a non-tech person, what are the minimum competences required to work at a data science business unit? Where should I start learning? — Diego Terceros Arce

I would suggest focusing on fundamentals, such as statistics. It is a basic requirement for data science and you need to have confidence in how you understand the data and what it is trying to tell you. Find tools that can help you visualize data, especially if they are high dimensional. Then move on to modeling, and look at how well your models approximate the real system.