How I Tripled My Income With Data Science in 18 Months
After losing the Job, I taught myself data science and tripled my income.
Photo by Karolina Grabowska
Around 18 months ago, I lost my job due to the COVID-19 pandemic. I was working as a part-time tutor while in college. The money I got from tutoring was used to cover expenses like food, petrol, and my car.
After the government imposed lockdown restrictions on the entire country, I was unable to continue teaching. I couldn’t go to college either and had to study at home.
Although this seemed terrible at first, I realized that not going to university and work freed up a lot of my time.
I started looking into expanding my skill set during this time. After doing some research, I found a machine learning online course that seemed pretty interesting.
That was the first online course I’d ever completed.
After that, I spent most of my time building projects, learning to code, and getting online certifications.
Now — 18 months in, I have built multiple income streams with my knowledge in the field of data science and analytics.
1. Full-time Job
I first joined a company as a data science intern for some time and am now working there full time.
At first, I expected my work to comprise primarily of model-building.
However, once I joined, I realized that my work was only about 10% model-building. The rest of the time, my team and I were looking into new solutions we could create to solve business problems.
Often, these problems didn’t even require machine learning to be solved. The data solution could just comprise of business logic translated into a simple SQL query.
The work I do daily involves answering questions like:
- How do we use data to find information about Company A’s competitors?
- We have built a customer footfall prediction model. What are the business use cases we can identify to test this model? Does it work as well in a production environment as it does in a test environment?
- How can we continuously improve segmentation and performance for our clients? Are we able to infer real-life scenarios from the data available?
This is a very abstract description of the kind of work I do on a daily basis, but I want to emphasize that creating a data science solution doesn’t start and end with model building.
If you are an aspiring data scientist, I suggest acquiring some domain knowledge in the industry you want to work in.
2. Data Science Blogging
I write about my experience in the field of data science.
If I build a project at work, I find a similar dataset on Kaggle and replicate the analysis, and create a tutorial around it.
I initially started writing and posting data science tutorials to enhance my portfolio.
Writing articles about my work was a way for me to connect with other aspiring data scientists. It was also a way for me to showcase my ability to code and build ML models.
Initially, I never expected to be paid for my writing. I just thought it was a great way to enhance my data science portfolio.
Over the past year, however, what started out as a hobby started to generate revenue.
I am able to make passive income now by simply creating data-related tutorials, projects, and writing about my experiences.
As I started to build an online presence within the data science community, I started getting multiple freelance offers. I’ve built machine learning models for clients on a one-off basis, created competitor-analysis reports, and written data science articles.
When I initially thought of freelancing, I pictured having to compete and bid for projects on an online platform.
However, all of my freelance clients have reached out to me after reading my articles or taking a look at my portfolio projects.
A few months back, I built a clustering algorithm and posted a tutorial about it online. The next day, someone reached out to me, asking if I’d be interested in building a clustering model for their client.
Freelancing has equipped me with a lot of skills outside of the domain I usually work in.
In my company, the data I work with usually comes in a certain pre-processed format, and I query the data with SQL and Python to make use of it.
When freelancing, however, client data comes in very different formats. Most of it isn’t processed or structured, and I’ve spent a lot of time figuring out the relationships between the datasets and making sense of it.
I also need to collect external data to come up with an analysis, and this usually involves scraping third-party websites and using open-source tools.
I feel like freelancing has given me exposure to knowledge I don’t currently have at my day job, and I’m able to learn new things with every project I take on.
How did I get here?
I mentioned above that I took a data science online course, and things changed from there. You might be wondering how.
To be completely honest, after taking my first data science online course, I felt lost. I spent about a month learning the different algorithms and training models with Scikit-Learn.
I simply didn’t know where to go from there.
I started reading articles about people who managed to land a data science job without a Master’s degree or any professional qualification. I realized the importance of domain knowledge and solving problems with the help of data available.
It wasn’t necessary for me to build the most accurate models or understand the underlying algorithm behind the model.
I realized that the most important skill for me to have was the ability to solve problems using data. This meant that I had to go beyond machine learning algorithms.
I took courses in business analytics and ML engineering. I spent more time learning to code than I spent on theory. I spent time learning SQL and data manipulation.
Then, I collected my own data from online sites with the help of web scraping. I used the data to solve a problem and built a simple machine learning web app with it.
This way, I slowly gained the skills required to become an end-to-end data scientist.
Even within the data analytics team at my job, if there are any projects that go beyond the scope of our daily work (projects that require external data collection or a new algorithm), I’m the one who usually gets assigned to it.
As an aspiring data scientist, there are so many resources made available to you online. Too many, in fact, that you don’t know what to choose from.
Most of the emphasis, however, is placed around model building.
While it is important to know the fundamentals of building and training a model, most jobs available out there require you to go beyond this.
The real demand is for people who can solve problems with the help of available data.