Today, I’ll cover one of the key factors that one should consider when thinking about a career in data science – Compensation. I looked at over 1,400 self-reported salaries from levels.fyi in order come up with the following breakdown:
- First Data Science job (0–3 YOE): $167k – $180k
- Mid-level roles (4–6 YOE): $204k – $220k
- Senior roles (7+ YOE): $240k – $265k
If you’re satisfied with the above, feel free to move on to whatever else you had planned in the day! But if you had five minutes to spare – I’ll go through some interesting insights I found from breaking out the compensation data, as well as some caveats to this self-reported dataset.
How am I defining compensation?
Typically, a data scientist’s salary will consist of three key parts:
- Base salary: Guaranteed cash paid out once or twice a month. Usually, the largest part of total compensation.
- Equity: Usually comes in form of RSU or options. Can be a large portion of compensation for more senior data scientists or for earlier stage start ups.
- Bonus: Either signing bonus, or performance related bonus (i.e. 10% of salary once each year)
For this article, I am defining compensation / total compensation to be the sum of the parts three above.
What dataset am I using?
I’m using levels.fyi, a place that collects user-reported compensation across many different roles in Tech. As of this article date (November 2020), there were 1,400 self-reported data points that I used for this analysis. The screenshot below shows the type of data that is available alongside compensation – we’ll look at each of these categories in more detail in the following sections.
For this article, I collected and cleaned the data using R (source code here). I recommend you play around this website for yourself, especially if you’re reading this article a few months after it was published!
Where are these roles located?
Out of the 1,400 self-reported data points, 90% of the roles were based within the United States. Because the data for outside-of-US roles is so sparse, I’ll focus this article on US-based roles only.
Geography-wise, even within the US, we see that a three-fourths of the roles are focused in just three major tech hubs:
- San Francisco / Silicon Valley (41% of responses)
Not too surprising, as this is where most of the major tech companies are headquartered or have a large office in. Google, Facebook, Apple, Netflix, and Uber are all headquartered here.
2. Seattle metro area (24% of responses)
Contains Microsoft and Amazon, which make up a majority of the responses here.
3. New York (10% of responses)
Most larger tech companies will have a sizable office here, so having NYC on this list is not too surprising.
If we wanted to break this out by individual cities, we have the graph below – outside of the major tech hubs, we see that several other major metro areas appear.
What companies am I looking at?
This section goes hand-in-hand with the previous section. Large companies make up a disproportionate share of the data science positions, and the data below confirms that. In addition, most companies are tech companies – the only companies below that are not explicitly tech are Capital One, JPMorgan Chase, and Booz Allen Hamilton. 40% of the data comes from just 6 companies: Microsoft, Amazon, Facebook, Google, Apple, and Uber.
Another way to look at these companies is to look at the type of employees that are reporting their compensation at each of these places. If we look at the average data scientist at each of these companies, here is what they look like in terms of their years of experience as well as their compensation.
Some interesting callouts from the graph above:
- The two rideshare companies have a stellar pay to YOE ratio. They are paying $250k-$280k for someone with 3–4 years of experience.
- Netflix pays their data scientists the most, with a median of $430k total compensation for someone of 5–6 years of experience.
- The most common type of employee across the various companies is someone who has 4–5 years of experience and is earning $200k-$250k
How does compensation increase with experience?
Compensation increases gradually with experience – the graph below tells the complete story. But to summarize:
- First data science job (0–3 YOE): $167k – $180k Total Compensation
- Mid-level roles(4–6 YOE): $204k – $220k Total Compensation
- Senior roles (7+ YOE): $240k – $265k Total Compensation
To add another dimension to the above, let’s focus only on median compensation and each of its components. The general trend as one becomes more experienced is that equity and bonuses become a larger component of their compensation package.
Of course, the graphs above are only as good as the data underlying it. Below, we see how many people reported compensation info for each YOE bucket. The coverage here is decent – for some of the smallest buckets, we have over 50 respondents, and in most categories we have over 100.
Fun Fact: One interesting tidbit to point out here is that data science is a relatively new career path that came to prominence in the age of big data. Because the need for data literacy is going to more important than ever moving forward, I’d be curious to see how compensation rises or falls to the demand and supply of data scientists in the upcoming years.
What does compensation look like at Big Tech?
So we’ve talked about compensation as a function of experience, as well as the companies that dominate this list. The next logical step is to try to compare compensation at these key companies!
Before jumping into compensation at the major tech companies, I wanted to talk about levels. At tech companies, talking about compensation only makes sense if you’re talking about it in the context of levels. Levels are determined based on years of experience, interview performance, as well as previous roles/education. For each level, you’ll have a set of expectations as well as a compensation range.
At Facebook, their entry-level data science position would be for IC3. From this dataset of 11 responses, we can assume that Facebook paid $144k-$190k in total comp for a data scientist coming in as an IC3.
Now that we have this concept of leveling, the graph below shows how different levels compare at the six major tech companies. I think this is one of the most interesting insights from this entire article as we can normalize salaries across major companies using comparable years of experience. Two key callouts:
- The higher the position, the more variable the pay: If we look at the bottom left corner, pay for data scientists with 0–4 years of experience is relatively standard across the major tech companies. Towards the top right, we see how people with 9+ years of experience have a much larger difference in pay across companies.
- Data is sparse at the highest levels: In addition to pay being more variable at higher levels, there are also less people occupying higher level roles at tech companies.
These numbers look different than what I’ve seen elsewhere
You may have stumbled upon this article after you looked through a few other articles about this same topic. When I do my own Google search, I see a pretty big range of salaries ranging from $85k – $130k. The numbers in this article are a little higher, however there are several reasons for this:
- Total compensation > salary: Oftentimes, other resources focus solely on salaries, while this article looks at total compensation which also includes equity and bonuses. Especially for more senior roles, equity can be a very significant portion of total compensation.
- A VERY biased dataset: This dataset is not representative of the typical data scientist in the US. The data is biased in several ways. (1) The companies represented in this data are heavily biased by tech giants that will pay more money. (2) These tech roles are also concentrated in high cost-of-living areas, where a higher salary is used to offset higher costs. For example, a $100k salary in San Francisco is similar to earning $55k in Dallas. (3) Lastly, the types of people who will know about levels.fyi and take the effort to self-report their data might be different than the general population of data scientists. These people may be more career-focused or interested in compensation, and as a result this might correlated to higher earners.
- Self-reported data: People can always exaggerate their compensation if it’s self-reported. However, we mostly look at medians in this article, so we are able to ignore outliers that may occur as a result of self-reporting. The larger issue here is that people may think of "Years of Experience" differently. Some people may include education or non data science roles towards YOE, while others may not.
In other words, when we consider the above three callouts, a more accurate title for this article would actually be:
How much do data scientists make at Big Tech companies in major tech hubs (Silicon Valley, Seattle, NYC)?
But that title is too long… so I went with a shorter version.
Conclusion
Hopefully you found this article informative and interesting. When I first thought about pursuing a career in data science, there weren’t many good resources on compensation, so I would have really appreciated an article like this. Having previously worked at companies that encouraged salary transparency, I have the mindset that increased transparency around this touchy topic is a good thing. So this is my contribution towards that effort!
If you enjoyed this article, you may also enjoy a similar analysis I did for product manager compensation:
