Photo by the Author
The obvious Getting started
I work as a data scientist at a large technology company. You know, the kind of company that pays well, has flexible working hours, and an office that looks more like a trendy cafe than a workplace (We have plush sofas and bean bags). My job in this company is a Product data scientist.
Often, big tech companies like Google, Meta, and Amazon hire Product Science Scientists to help drive millions of dollars in revenue.
In fact, Faang companies primarily hire product data scientists as part of their core teams, and these employees are highly compensated, often making more money than traditional data scientists. This is because product data scientists work closely with business teams and make decisions that affect millions of users every day.
I believe that in the age of AI, product data scientist roles are more secure than traditional scientific work. This is because the closer you are to influencing major business decisions, the harder it is to replace. While AI can create predictive models with good accuracy, it can't convince a product VP to kill a feature, and it can't gain enough understanding of a particular product to flatter stakeholders.
But I digarress.
Click on this article to learn how to get ACE Science interviews at major tech companies, and I won't make you wait too long.
Here is what I will explain to you in this article:
- What I do as a product data scientist.
- How did I prepare you for this product data science role, and what makes product data science different from other, traditional science jobs.
- My 6-week preparation plan to ace this data science interview.
- A must-read if you want to become a product data scientist (whether you already have data skills or are a complete beginner).
The obvious What I do as a product data scientist
In simple words, I use analytical techniques to answer questions like:
- Should we make this new feature, and is it worth the investment?
- How much money can we make on this new product launch?
- How do we use data to get users to engage more with the products and services we offer?
- How can we get people to spend as much time on the app as possible?
The obvious How I prepared for a data science interview
// 1. Start with data science skills
As we learned at the beginning of this article, product data science roles are different from traditional data science roles. Before applying for this job, I had 2 years of work experience as a predictive data scientist at another company.
This means that I already have the following skills:
- Programming: I was comfortable with Python and used it for web scanning, data analysis, and visualization.
- Data analysis: I knew how to do Eda with tools like Powerbi and can tell stories with data.
- Machine learning: I can build, train, and test machine learning models. This includes simple reuse models, as well as advanced topics such as research time prediction.
If you don't have these skills, I recommend watching My YouTube video How to get the basic knowledge needed to become a data scientist.
The above skills are easy to learn and will take 4-6 months to acquire.
// 2. Additional capabilities of the Product Data Science Section
Product data science requires a different set of skills than traditional data science roles. You're not just building predictive models like a product data scientist; You have to understand every product in the environment and help decide what are the building blocks, what works well, and what to kill.
Here are additional skills I should learn as a product data scientist:
→ SQL
SQL is the primary language of the product data scientist. All this while (as a traditional data scientist), I was working on Python books, and these days I almost rewrite SQL queries.
To learn SQL, I do two things. First, I took This SQL course of data analysis. After that, I spent 3 weeks solving SQL problems LeetCode and Hackerrank.
This practice was enough to get me through the technical part of the interview.
→ Statistics for decision making
I already know math and have taken many lessons in it. But as a product data scientist, I had to learn this skill Statistics used. This means that I have to use a programming language to find the confidence interval of the characteristic.
If a feature (like adding a pop-up to the screen) led to more engagement with a certain engagement moment, I had to decide that the product was worth launching. I also needed to understand concepts such as how to choose the right sample size for testing to ensure that our results are not limited.
If these concepts resonate with you, I suggest taking them this is a free course in Unfenent Statistics by Udacity. This, once Udacicity Free A/B Program Google, helped me answer the math and product related questions for this post.
→ Bridging the gap between math and business
A big part of product analysis is fundamentally bridging the gap between math and business. You decide on a successful metric for a particular product, and if the product performs well, you open it. For example, if your success metric is click-through-rate (CTR), you could say something like:
“A 2% improvement in CTR leads to another $1.5m in revenue every year, so we should ship this feature.”
Of course, the above example is overly simplistic, as product teams often generate many complex metrics to capture various aspects of user engagement.
Questions related to meter design and business use cases were the most difficult to answer during the interview. To prepare for this, I passed This product analysis course on Coursera (although I didn't finish it).
The obvious My Information Communication Process: Key takeover
To summarize, my Data Science product scout tested me on the following skills:
- Timed SQL challenges.
- Experiment Design and Statistics: “How would you construct the sample size for this experiment, and how would you determine the duration of the experiment?”.
- Business and product information: “Our current metric captures the number of times it can't find the desired result on the first page of search results. However, is just browsing the metric to be taken as a 'true failure of failure'?”.
The questions and discussion questions I shared in this article helped me come up with a data science role. Having worked as a data scientist for many years, I have learned that product data scientists are essentially good data scientists who know how to work with data.
Since we work a lot with business teams to make decisions that directly affect the bottom line of the company, I believe that this role is very important with ERA where AI can manage the standard model and analysis. If you are thinking about becoming a data scientist, or even if you already are one, I highly suggest looking into the product data science approach.
Yes, this role is very competitive since these roles are offered mainly by large technology companies and product organizations. However, if you invest the time and effort to prepare for a role like this, it puts you at the center of important business decisions, leading to more security and job security.
Natasha Selvaraj is a self-taught data scientist with a passion for writing. Natassha writes on everything related to science, related to the real king of all data topics. You can connect with him on LinkedIn or check out his YouTube channel.
