Data Scientist Salary in San Francisco — 2026 BLS Data
Salary distribution
Percentile breakdown of Data Scientist base salaries in San Francisco.
The $205K median base for a data scientist in San Francisco — derived from BLS OEWS May 2024 data for SOC code 15-2051 applied to the SF-Oakland-Hayward MSA — sits about $50K above the national median of $156K for the same metro pull and roughly $96K above the U.S. figure of $108,660. That gap is real, but the median masks a brutal bimodal distribution. Levels.fyi pegs the SF Bay Area band at $185K to $343K with a median total comp of $250K, and the top end of the city is currently dominated by a handful of AI labs paying packages most product DS roles will never touch. Treat the $205K base as a midpoint, not a target.
How SF DS salaries compare across cities
San Francisco still pays the highest data scientist base wages in the U.S., but the lead has narrowed since 2022 and the composition of the lead has shifted from “product analytics shops at FAANG” to “frontier AI labs.” SF median base sits at $205K. New York runs $180K–$195K, with the long tail driven by hedge funds (Two Sigma, Citadel, Jane Street) hiring DS-coded quant research roles that look more like ML engineering than experimentation work. Seattle lands $170K–$185K, anchored by Amazon’s measurement and pricing science orgs and Microsoft’s applied science teams.
Austin sits noticeably lower at $145K–$160K median. The big-tech outposts there (Indeed, Atlassian, Meta) pay close to bay numbers, but the local market average is dragged down by enterprise SaaS and Dell-orbit analytics roles that pay $130K–$155K. Remote-US bands at scale-ups typically land $155K–$190K — Anthropic, Stripe, GitLab, and Dropbox anchor the high end; a series-B startup hiring its first analytics DS lands closer to $145K–$165K.
What’s worth flagging: the SF premium is mostly captured at P75 and above, where AI-lab and infra-AI equity sits. A P25 SF role at $155K is essentially flat against a P50 NYC or Seattle role once you factor in cost of living. The city only “wins” the comparison if you can land in the upper half of its distribution.
What drives the spread in SF
The P25-to-P90 spread for SF data scientists ($155K to $440K base, and far wider on total comp) comes from three structural forces:
DS vs MLE pay gap. At the same company and level, machine learning engineer roles consistently outpay data scientist roles by 15–30%. A senior product DS at Meta or Airbnb lands $260K–$320K total comp; a senior MLE at the same company lands $340K–$420K. Title matters a lot here — “Data Scientist, ML” or “Research Scientist” often sit in the MLE band, while “Data Scientist, Analytics” or “Data Scientist, Product” sit in the lower band. Get the right code in your offer letter, not just the job description.
AI-lab premium. Anthropic, OpenAI, and a handful of well-funded labs (Mistral SF, xAI, Inflection-descendants) now pay research-coded DS and applied scientist roles $400K–$900K total at senior+, with equity that has appreciated 3–10x over the last two years. This single bucket of ~3,000 SF roles distorts the entire P90 figure. Outside the lab cluster, a P90 SF data scientist clears around $360K — still high, but a different universe from the lab numbers.
A/B-test-heavy product cos. Companies whose product moves on experimentation (Meta, Airbnb, Uber, DoorDash, Pinterest, Robinhood) pay product DS roles at parity with software engineering — same level ladder, same equity bands. Companies that treat DS as a reporting function (most enterprise SaaS) pay 25–40% less. Knowing which org you’re interviewing into matters more than the company logo.
Total comp: base + bonus + equity
For a typical L4–L5 product data scientist in San Francisco at a public tech company, the package looks roughly like this:
- Base salary: $205K. This is the BLS-tracked number and what lands on W-2 line 1. Internal bands are usually narrow — recruiters have ±5–8% flex.
- Target bonus: ~$38K. Public tech pays 15–20% of base as an annual cash bonus tied to company performance and individual rating. FAANG and Airbnb hit targets reliably; smaller publics underdeliver in down years.
- Annualized equity: ~$55K. Four-year initial RSU grant divided by four, valued at grant-date price. Initial grants at L4 typically run $180K–$240K, vesting 25/25/25/25 or front-loaded 33/33/22/12.
Total: roughly $298K at a mainstream L4–L5 product DS seat. Refresh grants start vesting in year 2 and add another $30K–$70K annualized once they kick in.
At L6 staff data scientist: base climbs to $230K–$260K, bonus to $45K–$60K, and equity dominates the package at $150K–$260K annualized. Total: $425K–$580K.
At an AI lab, the structure inverts. Senior research scientist base is often capped around $230K–$280K, but equity grants are $300K–$700K+ annualized at grant-date value, and the secondary tender prices typically run 2–4x the grant-date strike. Total comp at senior research level lands $550K–$900K, with the equity portion carrying most of the variance and most of the risk.
Signing bonuses for DS roles in SF run $20K–$60K at L4–L5 and $60K–$130K at L6+. Usually 12-month clawback.
COL-adjusted
San Francisco’s cost-of-living index sits at 178.6 — about 79% above the U.S. baseline of 100 and roughly 50% above Austin’s 119.3. That recalibrates the headline comparison.
A $205K SF median base has roughly the same purchasing power as $137K in Austin, $165K in Seattle, or $180K in New York. Run the comparison on total comp and the gap holds: $298K SF total comp ≈ $200K Austin, $241K Seattle, $263K NYC. The SF premium is real on a national-baseline basis, but most of it gets eaten by rent and state income tax.
The break-even point flips when equity vests well. An L5 product DS at a public company who gets 4 years of refresh grants and a 25% stock appreciation lands materially ahead of any other U.S. city. At an AI lab where equity has 5–10x’d, the COL math becomes irrelevant — those packages clear $700K+ and would still dominate after any reasonable adjustment.
Negotiation playbook
A few moves consistently move the needle for SF data scientist offers:
Anchor on total comp, not base. Recruiters open with base because it’s the band they have least flex on. Push the conversation to total comp, where equity refreshers and sign-on are far more negotiable. Phrase it as “I’m targeting $310K total” rather than “Can you bump base to $220K.”
Get a competing offer from a tier-up company. A Stripe or Anthropic competing offer at $340K reliably gets a Meta or Airbnb recruiter to come back $30K–$60K higher. A non-tier-up competing offer (e.g., a series-C startup) rarely moves the needle.
Negotiate the level, then the comp. Going from L4 to L5 changes the band by $60K–$100K total comp. Going from $215K to $230K base inside L4 changes it by $15K. If the level decision is even slightly in play, fight that fight before touching the numbers.
Reframe the equity discussion in dollars, not percent. A startup recruiter offering “0.3% equity” is much easier to negotiate with when you’ve translated it to a 4-year annualized dollar figure at the last-round preferred price. Always ask for the recent 409A and the last round’s price.
Use the BLS percentiles as a floor reference, not a target. If a recruiter pushes back on a $220K base ask citing “market,” the P75 BLS figure of $290K for SF DS is your reference point — quote it directly. It’s a federal data source they can’t dismiss as a self-selected board.
Caveats
The BLS OEWS May 2024 release is the most recent federal data, published in late 2024 and reflecting wages collected through a six-quarter panel ending in 2024. Real 2026 numbers for SF are higher — likely 5–10% on base, more on equity at AI-adjacent companies — because the underlying market has moved. Treat the percentiles as a directional floor, not the current band.
The BLS SOC code 15-2051 (“Data Scientists”) was created in 2018 and is still inconsistently coded across employers. Some companies file ML engineers under it; some file analytics engineers under it; some file research scientists under it. That coding noise explains a chunk of the wide percentile spread independent of any real pay variation.
Levels.fyi data is self-reported and skews toward FAANG and large pre-IPO companies. It underweights small-to-mid enterprise SaaS, biotech, and consultancies — which collectively employ most SF data scientists and pay below the Levels.fyi medians. If your target employer isn’t on the Levels.fyi top-100 list, expect actual offers $20K–$50K below the published medians.
Finally, the AI-lab pay tier is a moment-in-time phenomenon. Equity-driven outcomes at Anthropic and OpenAI have been historic, but they depend on continued funding-round step-ups. Plan around base + bonus when modeling your own number, and treat the lab equity as upside, not a load-bearing assumption.