I Spent $96k To Become a Data Scientist. Here Are 5 Crucial Lessons All Beginners Must Know
I spent so much money because I came from a business background, so I knew NOTHING about data science.
If this is also the case for you and you're feeling like you have no clue what you're getting yourself into, this article is for you (and my past self)!
I'd also be feeling clueless if I saw AI popping up new data science subfields each day. But don't worry! I got you.
I'm here to give you the heads-up I wish someone had given me 5 years ago when I was just a beginner.
Today, I'm sharing 5 crucial lessons that I learned from 3 years of data science training at top schools (including NYU), and 3 years of working at Spotify – 5 lessons that any data science beginner should know as early as possible!
I guarantee this article will help you better plan your own data science journey and fast-track your way to your career goals without following the same costly timely path.
You'll come out with a much better idea of what it means to be a data scientist today.
Lesson #1. Understand the Different Career Pathways of Data Science
When I started my data science journey, the field looked completely different from what it is today.
Data science sits at the heart of AI, so naturally, the field has been impacted by the same magnitude of change.
When you plan your data science education, you must take into consideration the different shifts happening within the data science profession under the influence of AI.
It starts with understanding what career pathways are available to you and finding the one that fits you best. Depending on your choice, you'll be managing your data science education differently.
Here's an example – What do you think is the difference between data science researchers and data scientists working in big tech?
The length and depth of their data science education.
The list of data science-related Careers is stretching by the day. But overall, they all fall within one of the following categories, and depending on your goals, you will have to draw your focus on one of them:
Category 1: Data Scientist Careers
- Data Scientists: They focus on analyzing data to extract insights, predict trends, and inform strategic decisions. This career requires skills in machine learning, statistical analysis, and data visualization.
- Decision Scientists: Similar to data scientists but with a stronger focus on decision-making processes. They use data insights to influence business strategies and outcomes. That's more the type of data scientist I am at Spotify.
- Business Analysts: They analyze data to understand business situations and propose measures for improvement. It's less technical than what data scientists do and with a stronger emphasis on business acumen.
- Quantitative Analysts: They specialize in quantitative data to solve financial and risk management problems. They usually develop statistical and mathematical models to inform strategies in finance.
Data scientist careers all have the same training foundation: → Strong foundation in mathematics (especially statistics). → Proficiency in programming languages (Python or R, SQL). → Expertise in data manipulation and analysis techniques.
However, the application of these skills varies:→ Data scientists and decision scientists are more focused on using data to inform broader strategic decisions. → Business analysts and quantitative analysts apply their skills more directly to business or financial contexts.
Category 2: Machine Learning Specialists
- Machine Learning Engineers (MLE): They design and deploy machine learning applications and systems.
- Natural Language Processing (NLP) Specialists: They develop models to process and understand human language.
- Computer Vision Specialists: They work on enabling computers to understand and interpret visual information.
- Large Language Models (LLM) Specialists: They **** specialize in developing and refining large language models for applications like chatbots and automated writing tools.
Machine learning-related careers all have the same training foundation: → Advanced degrees in computer science, Data Science, engineering, or mathematics. → Deep knowledge of machine learning algorithms, neural networks, and programming (Python or R, and sometimes C++ for system-level integration). → Specialization often (but not always) requires focused research, sometimes up to the doctoral level.
However, the application of these skills varies. The core training is similar, but specialists focus on different aspects of Machine Learning.
Category 3: Research Scientists in Data Science
I'd say they're all more or less the same. Regardless of what the core of their work is about, they all have the same foundational background:
- Typically a PhD in data science, machine learning, computer science, statistics, or some other related field.
- Stellar background in maths, especially in statistics and linear algebra.
- Deep expertise in specific research areas within data science or mathematics like audio systems.
The only difference here is their area of focus, similar to how machine learning specialists specialize too.
So if that's more the type of data scientist you'd like to become, make sure to adapt your education accordingly by making maths and computer science a core focus in your training. Be ready to commit to a PhD if you're aiming for the big Tech roles.
Lesson #2. Choose the Right Training Program
I jumped into data science when I started my master's degree at NYU four years ago.
It wasn't the best degree for the skill set I had at that time because 1. I didn't know how to code, 2. knowing some maths was far from being enough.
From the very first day, I was thrown right away into machine learning (ML). My coding skills were so basic that I struggled to write even two lines of code, let alone grasp complex ML algorithms.
I'm glad I pursued this degree, but a more strategic approach to how I prepared for it could have saved me a whole lot of mental and emotional sparring.
Not all programs are one-size-fits-all
If you're a beginner, it means you'll be coming with a different package of skills compared to other beginners. So you have to identify the programs that match your current skillset.
Here are some steps you can follow to achieve that:
- Evaluate the gaps in your background. Coming from a business background, I underestimated the importance of having foundational skills in coding and maths. It's essential to assess your existing skill set and identify any gaps in your expertise before jumping into new waters.
- Choose programs wisely. Opt for degrees or bootcamps that align with your current skills. These should ideally offer foundational courses in coding and maths before tackling advanced subjects like ML.
- Do your homework. Investigate the prerequisites of your chosen program thoroughly. If foundational courses are not included, invest time in developing these skills before the program begins. Don't start this program if you don't have the basics in check!
- Bridge the gap with online courses. If the program doesn't offer foundational courses, use the period before the program starts to fill any knowledge gaps. The initial two weeks of my second data science degree focused on refresher courses in probability, statistics, and linear algebra. It helped lay the grounds for what came next.
School prestige can matter depending on where you're located
I completed my first degree at NYU, a big name in the league of top unis in the U.S. Yet, that name didn't help much when I was applying for internships and jobs in America.
It's nice because they don't care as much about those things out there. What matters is whether you're skilled or not.
But in Europe, it's a bit different. I managed to pursue my second data science degree in the best engineering and business schools in France – Ecole Polytechnique and HEC Paris, and here, those names hold power.
They can make the difference between getting a second look at your resume or falling into the abyss with the others.
My big advice when searching for programs
If you're applying to a top school in the U.S. thinking a big name will help you land the big jobs, you may be wrong. However, it's always good to know that these can give you an edge in other parts of the world.
Keep that in mind when you're going for a degree. Adjust your expectations of what that degree will give you as a return on your investment later.
_Is it prestige? Is it the alumni network? Or maybe the degree is unique?_Do your research. Those things matter more than you think when you're an early-career. Unfortunately (or fortunately?)
Lesson #3. Learn Maths First. Not Machine Learning
Unpopular opinion but I think machine learning is overhyped. It's not the core of the data science job, mathematics is.
Machine learning can only get you so far in the field.
Most data science jobs belong to the first category I mentioned earlier, that of data scientists, and in that group:
ML is only used as a tool to answer business or strategic questions, not the answer itself.
If your goal is to land a job in a top-tier tech company, studying maths is a first & a must. ML alone won't get you far.
Actually, it won't get you anywhere.
Maths will.
A common mistake I see among most beginners is to focus directly on ML, but keep in mind that:
Data Science ≠ Machine Learning
ML as Product vs. ML as a Tool
I spent all of my data science studies developing all sorts of complex ML models. But when I hit the tech world, I barely dipped my toes in any of those ML sauces.
The truth is most data problems can be solved without ML
It took me time to understand this.
In the business world, ML has two use cases and people easily confuse the two.
As a beginner, you must internalize the fact that data scientists mostly use ML as a tool and not a product:
- ML as a Product – Recommender systems are a pure example of ML being the product or feature. Because the type of ML needed for this is so advanced and state-of-the-art, it's usually done by researchers with PhDs and massive maths luggage on their backs.
- ML as a Tool – Models are developed to help understand business questions. I spent 7 months developing an ML model at Spotify to identify the key factors driving listener satisfaction. In this case, ML is just a tool to answer business questions. It's a means to an end, not the end itself.
So focusing on ML itself will not guarantee you a seat in the company of your dreams because it's a nice-to-have only, not a must-have.
I know many data scientists at top-tier companies who are not particularly knowledgeable about ML. But can you guess what's the core skill they all master without any exception?
MATHS!
Here's why prioritizing maths over ML will give you an edge:
- A deep understanding of mathematical concepts is crucial. Without it, advancing in ML or AI is nearly impossible. You can learn any ML algorithm if you know the maths, but not the other way around.
- Stats over hype. Statistical knowledge is almost always more applicable and more valuable than ML in most data science roles.
- Top tech companies value maths expertise. Most business problems can be solved with statistical manipulation using methods like A/B testing for instance. Focus on those skills.
All in all, knowing ML will always be a good plus, but proficiency in maths, especially statistics, remains the fundamental backbone of data science.
Maths is the gift that keeps on giving. Once you put in the effort, you keep reaping the fruits continuously. Don't forget that.
Lesson #4. Business Projects Will Make You Standout
On my first day at Spotify, I learned a coworker had handpicked my resume. Naturally, I asked why mine stood out.
He said it was because I had a mix of business know-how and technical expertise.
He explained how this combo is what sets apart the best data scientists. All the ones he knew wore this powerful double-sided hat.
The lesson here? Focus on developing business acumen at all costs.
Here's why business insight is so critical
Most data science jobs require business expertise.
After working for 3 years as a data scientist, I understand now how being a good communicator is a non-negotiable when working in high-stakes environments.
At leading tech companies like Spotify or FAANG, data scientists are seen as key players in achieving business goals. So having the skills to reconcile business objectives and technical expertise will be your winning bet.
Where there's business, there's communication
Communication is the master skill for all business-related careers. It's what distinguishes leaders in any business field.
Being able to communicate complex data insights clearly and persuasively to stakeholders is invaluable. You'll have to convince decision-makers why your findings matter and to take action based on them.
It's this ability to communicate that amplifies your impact. It's the difference between a data scientist who merely analyzes data and one who drives real change.
It's the core skill that transforms good data scientists into great ones, that makes the whole difference between those who join top companies and those who don't.
Kaggle competitions won't help you much
Here's a secret: I've never done a real Kaggle competition in my life. I've always wanted to, but laziness got the best of me.
These competitions are good for getting your hands dirty but they won't help you much in landing a great job, the business projects will.
Here's how you can boost your business skills:
- Pursue projects with business data. The Titanic challenge is fun but prepares you in no way for the responsibilities of the job. Hands-on projects involving business problems are mandatory. Organize your findings in decks, and simulate a presentation to a non-technical audience to practice simplifying complex information.
- Pursue human-facing experiences. Participate in activities that require presenting to diverse audiences, such as meetups, hackathons, and workshops. This hones public speaking skills and the ability to make complex concepts accessible, crucial for roles in companies like FAANG and Spotify where explaining data science to non-experts is key.
- Real-world experience can give you an edge in the job market. Ideally, seek out internships or job experiences where you can gain the skills directly on the field.
One last thing, make sure those are the projects you highlight the most on your resume, especially for jobs belonging to the data scientist careers.
Lesson #5. The Only Way to Standout Is to Network
The job market is savage for everyone, but especially for new joiners. Getting a job today has become more challenging because the field is crawling with new joiners riding the AI hype.
Data science is no longer the heaven it used to be where companies spread their arms wide open to anyone who could turn a CSV into a dataframe.
So if you're thinking of starting your journey in data science, you have to be prepared to face the ferocity of the job market.
Even after graduating from the top schools in France, securing a good internship still felt like I needed to rub the genie out of his lamp to make it happen.
The year I got my internship at Spotify, I remember all of my class applying and getting rejected from internships at Ubisoft, including me. Buuut, I still managed to get the internship.
I networked my way into it.
Want to know how I did that? Here's a full guide to networking based on that:
Networking Got Me Hired in Tech Even After I Got Rejected, Here's How I Did It
Networking is no longer optional, it's mandatory!!
It doesn't take $100k to learn that networking will get you farther in life than the name of your alma mater – or pressing the send button on job portals.
The harsh truth is that less qualified people will get the job over you because they embrace the discomfort of networking. You have to learn to forge relationships if you want to achieve your career goals.
I used to think I could still land opportunities by trying hard enough.
But trying hard enough is no longer enough. You're dealing with dependencies that lie beyond your control.
Getting a call back relies first and foremost on luck. That includes everything from whether your resume landed on the right side of the pile to what the recruiter had for breakfast on that day.
Networking is a way to counteract luck by widening the net you cast. One fish is bound to take the bait after all.
The lesson here? Networking is now a crucial element of any winning job search strategy. You have to adjust to this facet of the job market because the rules of the game are set anyway.
Networking is the only way to stand out in a pool of extremely capable applicants. And I happen to have crafted a guide on how to create your own winning job strategy. So be sure to check it out thoroughly and digest its insides.
Was it worth it?
Yes!
I invested more than $100k in my data science education and 3 years of my life to learn the trade, and I would do it again because I love the freedom this job offers me.
But like I said, I could have managed things better had I learned sooner all of the lessons I just explained to you today. I'd have struggled less in my learning journey, and I would have adjusted my expectations and planned better for my career.
The good news is you don't have to spend $100k to succeed in your career, just imagine I did it for you. So that means all you have left to do is be mindful of how you approach and manage your education in the future.
How?
- Make sure you've understood the different career pathways available to you. Identify the one that aligns best with your interests and circumstances.
- Then research the different programs out there and pick the one that matches the most your current skillset. Don't forget to take foundational courses to fill any gaps in coding and maths.
- Prioritize getting the fundamentals in maths in check, don't rush into ML. It's a recipe for disaster and failure. I speak from experience.
- Business projects are the underrated GOATs in the job market. Make sure you have done a bunch of business projects from real-life scenarios to prepare you for the real responsibilities of the job. Kaggle is not your best friend unless it's serving business for dinner.
- Embrace the discomfort of networking because the job market is saturated with applicants. This helps you gain control over your own faith, don't let bad luck get the best of you. Take control!
