Data Science Salary Reality

Walk us through your compensation, concretely.

I'm going to give you real numbers because I think that's the only version of this conversation that's useful. My base salary is $182,000. My annual bonus target is 15%, which last year came in at 12.8%, so about $23,300. My RSU grant is $360,000 vesting over four years, which started two and a half years ago. Right now I'm vesting about $90,000 worth of stock per year, based on the grant price. The actual value fluctuates based on the stock price, obviously. At the current price it's worth about $112,000. Last quarter it was worth $96,000. My total comp if you add up base, bonus, and RSU at the current price is about $317,000. Some people's comp cards also include a signing bonus, which amortizes differently. Mine was $50,000 upfront when I joined, paid out over the first two years. So my first year total was higher.

What I actually take home is a different question. California income tax is 9.3% at my bracket. Federal is 32%. Medicare, Social Security. My net take-home on base and bonus combined, after maxing the 401k at $23,000 and a small pre-tax commuter benefit, is roughly $132,000 annually. The RSU adds to taxable income when it vests and is taxed at ordinary income rates at vest. So the RSU money runs through taxes at the same 32% federal rate. After tax on the RSU, I net maybe $62,000 more. All in, I take home somewhere around $194,000 net annually. Which is good. I want to be clear that I know that's very good.

Is it what you expected when you took the job?

I expected the number to be bigger in my head before I did the tax math. I grew up in a family where $200,000 gross household was "we've made it." And on paper I make $338,000. That number on a comp card looks like a different life. But $194K net is a real number, not an abstract one, and in the Bay Area, it feels different than it would in Columbus or Charlotte. My rent is $3,700 for a two-bedroom in Sunnyvale. My partner Camila's student loans are $890 a month, she did a design master's. Our car insurance is $340 per month for two cars. The cost of living adjustment is not a myth.

The thing nobody told me about tech comp is how the RSU cycle creates this weird psychological state where you feel rich on paper but anchored to the company. My next RSU refresh grant vests starting in about 18 months. If I leave before it fully vests, I leave money on the table. My manager Tariq, who's been at the company longer, calls this "golden handcuffs that feel like golden handcuffs." He's not wrong. I've been reached out to twice by other companies. Both times I ran the numbers and the unvested RSU made the move financially painful. So I've stayed.

What's the ceiling? Where do you go from L5?

L6 is the staff data scientist level. That's where total comp can hit $450,000 to $600,000 depending on the refresh cycle. The difference between L5 and L6 is scope of impact. At L5 you own a product area. At L6 you're driving technical direction across multiple areas or setting standards that other teams use. It's a meaningful jump in responsibility and a meaningful jump in comp. Most people who make L5 don't make L6. The promotion rate from L5 to L6 in my organization, based on talking to people who've been here longer, is probably somewhere around 20 to 25 percent. The rest stay at L5 until they leave. Which is fine. L5 is a very good place to be financially.

The thing about the ceiling that surprised me: it's not primarily about technical skill past a certain point. I know people at L6 who are good statisticians but not exceptional. What they're exceptional at is organizational influence, writing the document that changes the roadmap, running the working group that sets the ML standards, building the coalition of engineers who will actually implement what they specify. My friend Tariq keeps telling me "technical credibility gets you in the room, but it's the room skills that get you promoted." I don't love that. I'm better at the technical credibility part. The room skills part I'm actively working on.

What's yours?

How much of the comp depends on the stock. I don't control the stock. In 2022, the company's stock dropped about 35% in eight months. My unvested RSUs were worth about $95,000 less than they were at the start of the year. Nothing changed about my job or my performance or my output. The compensation just became less. The base and bonus didn't change. But the RSU, which is 35% of my total comp number on a good day, moved by $95,000 because of macroeconomic factors I had no part in. That's a real psychological effect that doesn't show up in the comp card. It's not unique to me. Everyone at my level has this. We just don't talk about it in the context of whether tech comp is "real" compensation. It's real when it's high. When it drops, you feel like you lost something even though your base hasn't changed.

Tell us the real numbers.

My salary is $108,000. There's no equity. The annual raise is typically 2.5 to 3.5%, tied to a performance review. My bonus is a 3% target, and it paid out at 2.8% last year, which was $2,900. Total compensation is $111,000. Benefits are good, which matters: healthcare is essentially free for me, dental and vision included, and the institute contributes 8% of my salary to a 403(b) retirement account regardless of whether I contribute, which is unusual and worth about $8,640 annually. Add that in and my total package is about $120,000. That's the honest version of the number.

Take-home in Ohio after federal, state, and FICA, maxing my 403(b) at $23,000, is about $68,000 net annually. My mortgage in Columbus is $1,340 a month on a house I bought in 2022. My student loan payment is $820 on the remaining $62,000 of my PhD loans. I'm comfortable. I'm not rich. I have enough and I don't feel anxious about money, which is something my colleague Jemma, who's a research associate making $58,000, cannot say.

You have a PhD in biostatistics from Ohio State. Have you run the numbers on what you'd make in tech?

Yes. A PhD with biostatistics background in the Bay Area would likely land at $170,000 to $200,000 base with significant equity. Total comp probably $280,000 to $340,000. That's not a secret. I've been recruited twice. The last one was about 16 months ago, a health tech company in San Francisco, the offer was $185,000 base and $400,000 in RSUs over four years. I spent two weeks thinking about it. I called my sister, who lives in San Jose. She said a two-bedroom apartment near her was $3,900. I looked at what I'd net after California taxes on that base. I did the housing math. After accounting for cost of living and what I'd take home versus what I take home now, the real purchasing power difference was about $40,000 a year in my favor in California. That's real. That's also the number that the mortgage on my Columbus house and the research I actually care about fit into.

The work I do here is on maternal health outcomes in rural Ohio. I work with a team of eight people including my manager Dr. Brennan, who is one of the best epidemiologists I've encountered, and a data engineer named Henry who keeps our entire pipeline running. The models I build inform state health department policy. Real policy, affecting real funding decisions that affect real people. In San Francisco I would be building churn models for a company that sells appointment-scheduling software to small clinics. That's a fine thing to build. It's not this thing.

That's the mission trade-off. But is there a ceiling problem?

Yes. The honest answer is yes. The highest individual contributor salary for a data scientist at this institute is somewhere around $125,000, and that's the director of analytics, which is technically a management role. As a senior individual contributor, I'm at $108,000. The next salary step is $115,000 to $120,000, which would require either a promotion to a role that doesn't currently exist or a renegotiation. I've had one salary renegotiation in six years and it got me $7,000. Dr. Brennan supported it. The institute's budget is grant-funded, which means it's not discretionary, it's tied to what grants come in and how they're allocated.

What I've been thinking about lately is whether at 38, with the research experience and the PhD, I'm underusing my earning potential in a way that will matter at 55. My financial advisor, a woman named Carol I've been working with for three years, has run the compounding math on the salary gap. The gap between what I'd earn in tech and what I earn here, over a 25-year career, compounded conservatively, is substantial. More than I'd like it to be. I'm not going to San Francisco. But I think about the number more than I'd like to.

What's yours?

The salary comparison culture in data science is really specific and relentless. Levels.fyi, blind threads, LinkedIn posts. It's not like this in other fields. Nurses don't have an anonymous forum where people post their total comp packages with stock breakdowns. I know exactly what I'm leaving on the table in a way that accountants at comparable organizations probably don't. That knowledge is a specific kind of tax. I made my choice deliberately and I'd make it again. But the visibility into the alternative is constant, and I have to actively choose not to look at it. The rule about Levels.fyi after 9 PM is real. It's how I keep the choice from corroding.

What do you make, and what does it look like in real life?

Base salary is $97,000. Bonus target is 8%, which last year paid out at 6.2%, so about $6,000. Total cash was $103,000. There's no equity. The company does have an employee stock purchase plan where I can buy shares at a 15% discount, and I've been doing that at the max allowed, which is 10% of my salary. So $9,700 a year pre-tax into ESPP, I buy shares at a discount, I sell quarterly and take the gain. Last year that netted me about $2,100 in profit, which isn't a huge number but it's literally free money if you turn it quickly. Total package including 401k match of 4% is about $113,000.

Take-home in North Carolina, no state tax credit surprises, maxing my 401k, is about $67,000 net annually. My rent is $1,290 for a one-bedroom in the South End neighborhood. I bought a used Honda in 2022, paid off. I save about $1,200 a month. I'm doing fine. I'm not anxious about money. I'm also not where I thought I'd be at 31 when I was in undergrad imagining what data scientist pay looked like.

Where did you think you'd be?

I read all the same articles everyone reads. "Data scientist: the sexiest job of the 21st century." The median salary figures. I had this impression that breaking into data science meant immediately being in the $130,000 to $150,000 range. What I found was that the high numbers are real but they're concentrated at specific companies. And getting to those companies is genuinely competitive. When I graduated, I applied to twenty data science roles and got six interviews and two offers. One was this job at $87,000 and one was at a startup in Raleigh at $84,000 with a tiny equity grant. I took this one for the stability.

At the time I thought I'd do two or three years and move into something bigger. I've been here three years and I'm now trying to figure out what that next step actually looks like. My manager Deon has been here nine years and he makes $118,000. The director of data and analytics, Renata, makes something in the $140,000 to $150,000 range, I'd guess. To get from where I am to where Renata is in this company, you're probably looking at 8 to 12 years. Or you leave for somewhere that pays more and takes you further faster. I'm in the process of figuring out which version of that I want to do.

What does "figuring it out" look like?

Mostly talking to people. I had coffee with a woman named June who was a DS at a mid-size consumer brand in Charlotte, then took a role at a healthcare tech company in Atlanta. She went from $102,000 to $134,000 with a small equity component. She said the two hardest parts were the technical interview bar, which was significantly harder at the new place, and the adjustment from being a generalist at a company that valued that to being a specialist at a company that wanted you to be deep on one specific problem. She said she'd do it again but it took her a year to feel competent. That's the useful part. People will tell you the salary number easily. Fewer people will tell you what the transition actually cost in terms of time and stress.

My friend from college Danny just got an offer from a tech company in Austin. He was at a similar company to mine in salary. His new offer is $148,000 base plus RSUs. We've been texting about it. He asked me if I was jealous. I said I was, a little, but I was also curious whether in two years he'd have the same answer about his new job that I have about mine. Which is: it's fine. I can do it without thinking too hard most days. I'm not sure he'll feel that way in Austin after 18 months of interview-loop stakes and performance review calibration. Maybe he will. Maybe the extra $51,000 a year is worth whatever that feels like. I honestly don't know yet.

What's yours?

How much the salary shapes what problems you work on. I'm at a retail company. I build demand forecasting models and customer segmentation models and the occasional churn model. These are not the most intellectually exciting problems in data science. They're important to the business, they're real, but they're not the kind of problems that stretch you in the way that working on recommendation systems at a company with a billion users would stretch you. And the skills you build in demand forecasting at a mid-size retailer are valuable, but they're not the same skills you'd build in that other environment. So the salary difference and the problem-type difference are actually two sides of the same coin. The companies that pay the most are often the ones working on the most interesting technical problems. You're not just trading salary for stability. You're trading a certain kind of intellectual development for stability and predictability. That's not a bad trade necessarily. But I don't think I fully understood that it was part of the trade when I took this job. I thought I'd just do data science and the interesting problems would find me wherever I was. That's not quite how it works.