The world’s leading publication for data science, AI, and ML professionals.

How To Advance In Data Science

My tips and experience to become a high quality data scientist

Photo by Tima Miroshnichenko: https://www.pexels.com/photo/man-in-white-dress-shirt-using-laptop-7567434/
Photo by Tima Miroshnichenko: https://www.pexels.com/photo/man-in-white-dress-shirt-using-laptop-7567434/

Once you become a data scientist, it’s a great feeling knowing that your days, weeks, and even months of studying and learning have finally paid off.

However, this is just the start.

Chances are you don’t want to be any data scientist, but probably a great one. So, in this article I want to go over several things you should do to push yourself into the top 1% of data scientists from my personal experiences and from what I have observed from these top echelon of practitioners.

Consistent Improvement

Photo by David Gavi on Unsplash
Photo by David Gavi on Unsplash

An idea that the book Atomic Habits (by James Clear) made popular is that improving 1% every day greatly compounds over time.

  • 1% improvement every day is 1.01³⁶⁵ ~ 38
  • 1% decline every day is 0.99⁹³⁶⁵ ~ 0.026
Diagram by author.
Diagram by author.

By making small positive choices each day as a data scientist, your progress will compound, pushing you into that elite tier.

Ok sure, this sounds nice on paper, but what can you do to implement this in practice?

Well, the best way is to learn something new each day. It doesn’t matter how small it is, if you are picking up knowledge every single day, needless to say, this will benefit you in the long term.

I can give you some examples to try:

  • Instead of Pandas, use a package like Polars or Spark to learn a new data manipulation framework.
  • Write aliases for any terminal or command line prompts you frequently use.
  • When doing a code review, ask why someone has done something a certain way.
  • Pair with a software engineer to learn some productivity tips for your IDE.

A good way to keep yourself accountable is to write down something new you learned at the end of the day.

Take Ownership

Photo by Markus Winkler on Unsplash
Photo by Markus Winkler on Unsplash

There is nothing wrong with doing your day-to-day tasks to a high level. However, to stand out and excel in your Data Science career, you need to start taking ownership of projects and products.

What this means, is that you take a proactive approach to improving current systems, processes, and models. Instead of waiting to be told what to do all the time, you actively look for areas you can make better.

Having this initiative will improve your problem-solving, leadership, and an array of other skills. Not to mention, it will probably catch the eyes of seniors and stakeholders within your company.

So, let’s give some examples of how you can do this.

These are mainly targeted at junior or mid-level data scientists, although I am sure the seniors among you can get some inspiration nonetheless!

  • When working on a codebase, if you find a bug or model improvement, mention it and implement it.
  • Automate any manual processes like getting certain data or model monitoring.
  • Volunteer to lead projects if possible doesn’t matter how small they are.
  • Research potential model improvements and bring them to planning sessions with some notes.
  • Organize things like hackathons that are centered around improving something within the company.

There is a range in how easy some of things are to do and it also depends on the structure and your position within a company. The main point is, to try and seize ownership of anything when you have the chance and deliver the requirement as best as you can.

Master Soft Skills

Photo by Harli Marten on Unsplash
Photo by Harli Marten on Unsplash

Even though data science is a very technical job, soft skills are essential to help you move up the ranks. All data scientists have a solid foundation in maths and coding, but if you can communicate and articulate clearly, your ideas will have more influence.

Being able to explain complex mathematical models like neural networks to non-tech-savvy stakeholders is truly a superpower. You are a translator.

In most companies, data scientists work in cross-functional teams full of engineers, product managers, and analysts. All these roles have varying levels of understanding of data science. If you can work smoothly within these teams, then it will aid in getting the work done more efficiently.

Perhaps one of the best benefits having great soft skills gives you is that of trust and influence. If people trust you, you can start influencing decision-making within your team and company. Needless to say, this is very important.

Improving soft skills can be slightly difficult as some are just part of human nature and way we naturally are. However, the following suggestions should be useful for most people:

  • Take opportunities to present to different audiences. Don’t just present to data scientists, show your work to non-technical people and tailor the presentation to them.
  • Make an effort to interact with people you don’t have frequent meetings with but are relevant or impacted by your work.
  • Contribute actively in meetings, don’t be another fly on the wall.

Developing soft skills is a life-long endeavour, but there are ways to expedite your progress.

Learn Adjacent Skills

Photo by Thought Catalog on Unsplash
Photo by Thought Catalog on Unsplash

Let’s say you are a data scientist who specializes in recommendation systems. It’s a great area to be in and there is demand for this skill.

But do you know what’s better?

Being a data scientist who specializes in recommendation systems and knows how to deploy their algorithms to production effectively using software engineering principles and cloud systems.

What I am talking about here is the idea of skill-stacking. Indeed defines skill-stacking as

"Skill stacking is the concept that individuals can make themselves more valuable by gaining a wide range of skills instead of pursuing one skill or talent"

In other words, learn new skills that crossover and complement your existing knowledge.

There are many great data scientists out there, but how many of them also know things like web development, MLOps, or software engineering to a good level?

It doesn’t even need to go that broad, you can learn another area in data science to a high level. You can be an expert in recommendation systems and say computer vision.

This crossover of skills makes you much more valuable to employers as you are one person capable of doing some bits of other roles.

How do you go about acquiring these adjacent skills?

  1. Identify an area that you want to learn about in which you do not currently don’t specialize.
  2. If possible find opportunities or projects in your job that allow you to work on said skill. For example, if you want to learn computer vision, there may be a computer vision project at your company that you can get involved in.
  3. Learn and study outside of your job and show your work online to demonstrate your understanding.

You basically want to find the easiest way to incorporate learning adjacent skills in your day-to-day life.

Put In More Time

Photo by Nathan Dumlao on Unsplash
Photo by Nathan Dumlao on Unsplash

To be honest, to become a top 1% data scientist, you have to put in more time honing skills than others are willing to do.

Nowadays, "hustle culture" is seen as a negative thing, and that’s not what I am insinuating. Don’t work non-stop all day trying to learn everything under the sun. That’s not practical or productive and will eventually lead to burnout.

However, there is something said about putting in that extra hour or two each day to refine your abilities and learn new ones. Again, small incremental gains compound over time.

How do you go about finding more time to develop your abilities?

Well, this can be a whole article in itself, but I can give you some suggestions that have worked for me in the past and I still use today:

  • If you work from home some days, use the time you would spend commuting to learn new things.
  • Wake up an hour earlier, this gives you another hour. Needless to say, don’t sacrifice your sleep for this!
  • Time block learning segments in your calendar to help encourage you to dedicate learning time.

It doesn’t matter when or how you dedicate this extra time, the point is that you have it and utilize it effectively!

Of course, you can become a great data scientist by simply performing well in your day job and spending your time outside hours doing other things. I want to re-iterate that there is nothing wrong with this and I am well aware that people have other pressing priorities, which makes learning outside of work difficult.

This is for people who want to be at the top of the top of data scientist and that simply means putting in more hours sometimes.

Do you think top researchers like Andrej Karpathy or Yann LeCun simply worked 9–5 to get to where they are today? I bet probably not.

Summary & Further Thoughts

You might have noticed a key theme within those five topics I just talked about. To be a top 1% data scientist you have to continually learn and invest more time than others. I am sure this comes as no surprise to you, but there is no shortcut to becoming truly great at a skill or profession just hard work and effort. It is important, however, to have direction and a smart approach to this extra learning time you are putting in. Working hard in the wrong direction can be dangerous, so you need to align yourself first. I hope this article gave you some insight into the things I have observed that the top data scientists do and can help propel you forward in your career!

Another Thing!

I have a free newsletter, Dishing the Data, where I share weekly tips for becoming a better Data Scientist. There is no "fluff" or "clickbait," just pure actionable insights from a practicing Data Scientist.

Dishing The Data | Egor Howell | Substack

Connect With Me!


Related Articles