We mapped de-duplicated open jobs across US-headquartered AI & tech startups to answer four questions a job seeker actually cares about: who is hiring at each stage, how hard they're hiring, where the jobs live, and — pulled from the raw posting text of JDs — what they actually pay.
Most "where should I work" advice collapses into vibes: Series A is fun, late stage is safer, join before the IPO. The data tells a sharper story. The funding stage of the company you're considering is the single biggest predictor of what kind of job is even on the table, how hard the company is hiring, how much capital is behind every existing employee, and — for roughly one in five postings — what the base salary range actually is.
A few honest caveats up front. Our tracked companies are curated — they are tech and AI startups we cover, not a representative sample of US startups. About have at least one open role in this snapshot; the rest are between hiring sprints. And the data is a single point-in-time snapshot, not a flow — when we say "Series D companies hire Sales", we mean the open reqs they're listing today. Salary numbers come from JD body text (mostly California postings forced to disclose under SB 1162), not from the structured ATS field; coverage varies by stage.
Every startup has the same headcount problem: too much to build, not enough people. But the shape of that problem rotates predictably with funding stage. Below is the share of open roles in each functional area, across stages with enough data to read (≥ 500 postings).
Engineering Owns Seed. Sales Owns the Middle. Late Stage Goes Back to Code.
Share of open postings by department · per-job analysis
Read At Seed, 63% of postings are engineering; sales is barely 15%. By Series D, engineering drops to its lowest (47%) and Sales/GTM peaks (29%). Series G+ snaps back to engineering-heavy — driven by AI-infra and fintech giants that never stop hiring builders.
—
—
—
Good news: engineering is the largest share at every stage. But two stages over-index hard: Seed (63% engineering, almost no GTM yet) and Series G+ (driven by AI-infra giants). Series D is when engineering drops to its lowest — these companies have shifted attention to selling what they already built.
You barely exist at Seed. The hiring window opens around Series B and peaks at Series D (29% of all postings). The brand is real enough to open enterprise doors, OTE pressure is highest, management is willing to pay top-of-band. Just understand you're being hired to justify the markup from the last round — most of the equity upside has already been priced in.
Operations roles are 6% of Seed postings and stay flat at 10–13% from Series A onward. Series D shows a quiet bump (12.5%) — that's when companies bring in "adult supervision": GC, VP Finance, senior people-ops. Late Series C through D is your inflection point.
Two companies both have "open roles". One has 12 employees and 8 reqs. The other has 800 employees and 8 reqs.
These are not the same job market. We can measure hiring intensity directly:
open_jobs / current_headcount.
Above 0.3 means the company is trying to grow by more than 30% in a single hiring cycle. Below 0.05 means they're in maintenance mode.
The Earlier the Stage, the More the Resume Actually Gets Read
Open jobs ÷ current headcount · per-company distribution
Read Median intensity drops monotonically from Seed (0.33) to Series D (0.16). Late stages have too few sampled companies in this dataset (cos < 30) to plot a stable distribution — but the direction is unmistakable.
Seed and Series A companies hire at median 0.24–0.33 intensity, with top performers above 1.0 (more open roles than current employees). Recruiters at these companies are desperate, response times are days not weeks, and you can negotiate hard. Trade-off: you'll often interview with the CEO who has no time, the process is chaotic, and some of these companies won't exist in 3 years.
Series D is the most interesting stage in this dataset and the one most likely to fool a job seeker. Two metrics peak here together: GTM share of postings (29%) and capital density — total funding raised per current employee (median —/head, the highest of any stage with enough data to read).
Sales Hiring and Capital Density Peak in the Same Stage
GTM % of postings (bars) and median capital density (line) · Seed → D
Read This is what a "scaling moment" looks like from the outside: maximum cash on the balance sheet per existing employee, and a sudden pivot toward salespeople. From the inside, it's also when growth pressure is at its highest — the valuation has been marked up aggressively, and Series E needs to justify it.
Gold rush
You're late
Capital density — total funding divided by current employees — is one of the only readable proxies for company-level financial health in this dataset. Healthy growth: density climbs as the company raises bigger rounds. Unhealthy: density flattens or drops, meaning the company is hiring faster than it's raising — burning runway per head.
Capital Per Head Climbs Roughly Monotonically Through Series D
Total funding ÷ employee midpoint, by stage · log scale
Read Median rises from Seed ($600k/head) to Series D ($1.34M/head). Stages beyond D have too few companies in this sample for the distribution to be reliable — we mention them in the methodology section as anecdotes only.
Money vs. People — One Dot Per Company
Total funding ($M) vs. employee count · dot size = open jobs
Read Most companies cluster on a $500k–$2M-per-head diagonal. Outliers above are "rich but small" (well-funded early stage); below are "lean and large" (bootstrapped or older private equity holdings).
Job titles are the cheapest negotiating chip a company has. How they hand them out tells you about the org's maturity. Below: % of postings at each stage whose title contains a given keyword.
"Founding" Vanishes Fast. "Staff" Explodes at Series E.
Title keyword frequency by stage
Read Brighter shading = higher share of postings containing that word. "Founding" titles peak at — of Seed postings and effectively vanish by Series D. "Staff Engineer" inflates dramatically at Series E (19.4%) — three times the Series B rate.
"Founding Engineer", "Founding PM", "Founding GTM" — these titles peak at Seed and effectively vanish by Series D. If "founding" is on your career bucket list, your window is Seed to early A. Real meaning: small enough team that you'll touch everything, but also pre-PMF risk.
"Staff" appears in 19.4% of Series E postings — more than 3× the rate at Series B. Cohort interpretation: at Series E, companies are competing for senior engineers against FAANG, and "Staff" is the cheapest title bump they can offer. If Staff title matters to your career, Series E is statistically the easiest place to get one.
SF Bay Owns Seed. Remote Peaks in the Middle. Late Stage Goes Everywhere.
Location distribution of postings by stage
Read SF Bay share drops from ~28% at Seed to ~14% by Series E. Remote share rises in the opposite direction, peaking ~20% at Series E. Late stages disperse most of their hiring outside the SF/NYC core.
Remote share rises monotonically from 10% (Seed) to a peak of 19.8% at Series E, then collapses at later stages. Mid-to-late companies are the remote-friendliest cohort — they've built distributed infra, but haven't yet ossified into "back to office" cultures.
We almost cut this section. The structured "compensation" field that ATS systems expose is filled in for 0.01% of postings — three jobs out of 35,000. A more responsible analyst would have stopped there.
But postings have bodies, not just structured fields. California's SB 1162 (in effect since 2023) forces
any company with 15+ employees to disclose a pay range in the posting itself if the role can be performed in CA. So we
walked every cached JD body we had (22,529 of them) with a regex for $X – $Y ranges, classified each match
as base / OTE / unclassified, sanity-bounded the values, and rescued
structured salary records —
job-level coverage. That's roughly the same coverage rate Levels.fyi achieves through self-reported submissions,
obtained from a regex.
Base Salary by Funding Stage — Median and Inter-Quartile Range
USD, midpoint of disclosed range · base-pay postings only (excludes OTE / hourly)
Read The shaded band is p25–p75 — half of disclosed roles fall inside it. Note the Series A trough at ~$150k median, lower than every neighboring stage. Series A is when companies hire mid-level talent at "standard market" rates; Seed inflates because it's dominated by AI labs paying frontier-research premiums (more on that below).
—
—
—
The Real Mapping — Median Base Pay by Stage × Department
Cells with fewer than 10 disclosed roles are left blank
Read Engineering pulls 20–40% premium over GTM / Operations at every stage. The richest cells: Series C and F Engineering at ~$213k median. The leanest: Series A Operations at $140k. Hover for cell-level sample sizes.
GTM Roles: Base vs. OTE
"Total comp" for sales roles is meaningfully different from the base-only number
Read Postings that explicitly mention "OTE" or "total compensation" cluster about — higher than ones that quote base salary. If you're a sales hire, your target number is the OTE distribution, not the base. Most companies disclose only one or the other — not both.
Base Pay by Company Size
Median + IQR, by employee count bucket · base pay only
Read Salary is not monotonic in company size. The dip at 11–50 ($148k median) reflects A-stage companies hiring junior/mid. Larger companies (200–5,000) hover around $180–200k. Single-digit-employee companies paying $200k+ is a Seed selection artifact — see next chart.
"Seed" Is Now Two Different Job Markets
Each dot is one Seed-stage company · X = total funding raised · Y = median disclosed base
Read Traditional Seed (left cluster) pays $100–150k. But the upper-right is the new phenomenon: a cohort of frontier AI labs raising $50M–$300M at "Seed" while paying Series E rates ($250–450k). The Seed label has lost its ability to predict comp on its own.
Don't ask the stage. Ask the last round amount. A "Seed" company that raised $5M and a "Seed" company that raised $300M are running different businesses, hiring different people, and paying differently — sometimes by a 3× factor. The stage label is no longer enough information.
Every claim above ("Seed = Engineering-heavy", "Sales peaks at Series D", "Engineering pays $20k more than GTM") was computed across all companies pooled together. But this dataset spans eight pretty different industries, and a pattern that holds for AI labs might collapse in healthcare or fintech. So we did the obvious robustness check: re-ran the headline analyses inside each industry bucket. Cells with fewer than 80 postings are left blank — we'd rather show a gap than a noisy number.
Engineering Share — Stage × Industry
% of all postings at each (industry, stage) that are engineering / product roles
Read The "Engineering owns every stage" finding is mostly real — but with two exceptions. FinTech and Enterprise SaaS companies invert at Series D onward: GTM & Operations grab a larger share than engineering. AI Infrastructure / Models stays the most engineering-heavy throughout.
GTM Share Across Stages — One Line Per Industry
Does the "Series D Sales Peak" hold inside every industry?
Read The headline "Series D Sales peak" we identified in §3 is true on average — but it conceals four very different industry trajectories. Cybersecurity hits the strongest D peak (43% GTM share). AI / ML matches the textbook curve (30% at D). Enterprise SaaS and DevTools / Data actually peak earlier at Series C (41% and 33%). FinTech / Crypto peaks earliest of all — Sales/GTM is already 29% by Series A and falls through Series D. Healthcare / Bio and Hardware / Climate have flatter GTM curves throughout.
The pooled Series D peak we showed earlier was partly an averaging artifact — the result of adding up eight industries with quite different GTM-hiring curves. The "D = Sales gold rush" framing only holds cleanly for AI / ML, Cybersecurity, and Hardware / Climate. For Enterprise SaaS and DevTools, the right target is Series C. For FinTech, you want to be looking at Series A or B — by Series D, FinTech companies have largely finished their early-sales hiring sprint and shifted to operations and management hires.
Median Base Salary — Stage × Industry
USD, base-pay postings only · cells with N < 15 disclosed roles left blank
Read AI Infrastructure / Models pays the highest median base at every stage ($200–250k where data is sufficient), confirming the "Seed inversion" earlier was an AI-cohort artifact, not a stage effect. Healthcare / Bio and Hardware / Climate have the lowest medians, consistent with the lower-cap pay markets in those sectors. FinTech and Enterprise SaaS land in the middle.
Stage alone is a weak predictor of pay. Industry × stage is far stronger. A Series A AI Infra role pays as much as a Series D FinTech role. If you're optimizing for cash, choose industry first, stage second.
The other variable that quietly shapes risk: how long since the company last closed a round. A "Series F" that just closed yesterday and a "Series F" that closed 3 years ago and hasn't raised since are living in different financial realities. The stage label hides this completely. We bucketed every company in our analysis by lag since most recent round: Fresh (< 12 mo), Recent (12–24 mo), Stale (> 24 mo).
How Fresh Is The Money? Lag-Since-Last-Round, By Stage
Per-company breakdown · stages with ≥ 20 sampled companies only
Read Earlier stages skew Fresh; later stages accumulate Stale. By Series D, nearly a third of companies haven't raised in over 24 months. These aren't dead — but their runway picture is meaningfully tighter than the "Series D" label suggests.
Red Flags — Stale Companies Still Hiring Hard
Companies with last round > 18 mo ago AND open roles > 5% of headcount AND ≥ 3 open jobs
Read These are companies whose hiring tempo is out of step with their funding tempo. Could mean (a) they're burning runway in a sprint to revenue before a Series X, (b) they're profitable and bootstrapped-on-old-equity, or (c) they're about to hit a wall. Dot size = open jobs; X = days since last round; Y = open jobs ÷ headcount. Hover for the company name. Always ask about runway when interviewing here.
Ask: How much runway do you have at current burn? If the answer is < 12 months and there's no revenue pivot in motion, you're joining what could be a layoff cycle within your first year. Companies named above include some great ones — but the question is non-negotiable in 2026.
Map what you want to which funding stage is statistically friendliest to that goal.
| If you want… | Target stage | Typical base pay | Why the data says so |
|---|---|---|---|
| Founding-title role | Seed → early A | $150–250k (high variance) | 4.4% of Seed postings have "Founding" — vanishes by D. |
| Maximum engineering pay | Series C or F | ~$214k median | Highest engineering-cell median in the mapping. Standardized comp bands. |
| Greenfield engineering | Series A → C | $165–214k base | Engineering share high; roles not yet specialized. |
| Sales / GTM gold rush | Series B → D | $160–180k base · OTE noticeably higher | GTM share peaks at 29%; OTE postings cluster above base. |
| Staff-Engineer title bump | Series E | ~$188k base | "Staff" appears in 19% of titles — 3× the Series B rate. |
| Remote-friendly role | Series C → E | ~$180–195k base | Remote share peaks ~20% in this band. |
| People-management track | Series D → F | $170–220k base | Manager / Director density rises with each round. |
| Frontier-AI / research role | "Seed" (verify funding ≥ $50M) | $250–450k base | The new Seed cohort — pays Series E rates if backed by mega-Seed. |
| Career stability over upside | Series F+ (case-by-case) | $140–210k base | Sample too small to generalize — verify each company. |
Data: Job postings collected from ATS systems
(Greenhouse, Lever, Ashby, Workable, Comeet, Gem, Consider) across
— US-headquartered AI & tech startups in the
Fast AI Startup Jobs dataset. Funding stage and employee count come from manually verified company
records; headcount uses the LinkedIn-reported bracket midpoint
(e.g., 51–200 → 125).
De-duplication: The raw jobs-lite.json snapshot
contained rows but only
unique posting keys
( duplicates removed). All charts in this article use the de-duplicated set.
Sample sizes: companies have at least one open role in this snapshot. of those have a recognized funding stage and drive the per-job charts (department mix, geography, title keywords). Capital density and hiring intensity charts use the companies that have funding + headcount + stage data. Stages with fewer than 30 companies (Pre-Seed, Series E onward for per-company charts) are excluded — their numbers in tooltips and stat boxes are directional, not statistically defensible.
Salary extraction: We walked every cached JD body
(22,529 files at build time) with a multi-pattern regex matching $X – $Y ranges
(including unicode dashes, "K"-suffix shorthand, and "to" connectors). Each match was sanity-bounded
(annual base ≥ $20k, ≤ $1.5M, ratio max/min ≤ 3.5×) and classified as base, OTE, or
unspecified based on contextual keywords ("base salary", "OTE", "total compensation").
salary records survived parsing — joinable to
unique postings.
Coverage skews toward California-based postings (where SB 1162 mandates disclosure since 2023).
Salary charts use base only unless explicitly noted otherwise — mixing base and OTE inflates
sales-role medians artificially.
What this analysis is not: a representative survey of the US startup market. The tracked company list is curated for AI & tech relevance. It is also a single-point snapshot, not a time series. We make no statistical inference claims — every comparison is a point estimate without confidence intervals. Bootstrap CIs and industry-segmentation robustness checks are planned follow-up work.