47% of AI Startup Jobs Aren't Engineering Roles. Here's the Data.
By Fast AI Startup Jobs
For every engineer at an AI startup, there's almost exactly one person in sales, marketing, finance, legal, or operations. That's not intuition. It's what the data says.
We analyzed 35,450 active job listings across 1,542 AI startups tracked in the Fast AI Startup Jobs dataset. Of those, 16,511 jobs, 46.6% of the total, fall outside engineering, research, and product entirely. The ratio of technical to non-technical hiring is nearly 1:1.
This matters because the prevailing story around AI careers is almost entirely about coding. Learn Python. Study machine learning. Get an MLOps certification. But the companies building AI products need accountants, salespeople, lawyers, recruiters, and operations leads in roughly the same numbers as they need engineers. The market is signaling something the career advice industry mostly ignores.
What the numbers show
Treating AI as an engineering discipline is understandable, since the underlying technology is deeply technical. But building a company around that technology is not. It needs the same commercial infrastructure as any other B2B software business: people who can sell it, market it, support customers, manage finance, and handle legal exposure.
Our dataset of 35,450 AI startup job listings breaks down as follows:
| Category | Jobs | % of Total |
|---|---|---|
| Engineering, Research & Product | 18,939 | 53.4% |
| Non-Engineering (all categories) | 16,511 | 46.6% |
| Total | 35,450 | 100% |
The gap between technical and non-technical hiring is 6.8 percentage points, close to parity. A 53/47 split is about the same as a coin flip with a slight bias. These are not edge cases or support roles bolted onto an otherwise engineering-heavy organization. They are half the company.
What non-engineering roles actually look like
The 16,511 non-engineering jobs spread across eleven distinct function areas:
| Department | Jobs | % of All AI Jobs |
|---|---|---|
| Sales & Partnerships | 4,991 | 14.1% |
| General / Unclassified | 4,791 | 13.5% |
| Marketing & Growth | 1,639 | 4.6% |
| Customer Success & Support | 1,228 | 3.5% |
| Finance & Accounting | 1,208 | 3.4% |
| Legal | 949 | 2.7% |
| Recruiting & Talent | 622 | 1.8% |
| IT & Internal Systems | 393 | 1.1% |
| HR & People | 296 | 0.8% |
| BizOps & Program Ops | 270 | 0.8% |
| Executive Leadership | 96 | 0.3% |
| Total | 16,511 | 46.6% |
The "General / Unclassified" bucket is large (4,791 jobs) because many startups post roles with non-standard titles that don't map cleanly to a single function. These include roles like "Founding Community Ops Lead," "Chief of Staff," or "Business Development — Strategic Partnerships." They are not engineering roles under a different name. Even if you excluded this entire unclassified bucket, the remaining non-engineering roles would still account for 33% of all AI startup jobs, a third of the market.
Sales is the dominant category by a wide margin, and that leads to the most counterintuitive finding in the dataset.
8x more sales jobs than ML engineers
The dataset holds 4,991 Sales & Partnerships jobs and 616 ML/AI Engineering jobs. Sales outpaces ML engineering by a factor of 8.1x.
That comparison pits the largest non-engineering subcategory against one of the smallest engineering ones. Looked at more broadly, Software Engineering as a whole has 11,241 listings, still 2.3x the size of Sales. Engineering is large. But Sales alone running 8x the size of the ML/AI specialty that defines the industry's public image tells you something about where headcount actually goes once a company has a product.
AI products are technically complex but commercially simple in one respect: someone still has to sell them. Enterprise AI tools take significant sales effort because the buying process is long, the procurement cycle pulls in legal and security review, and the ROI case has to be made to several stakeholders at once. That requires human salespeople, and a lot of them.
The ML/AI engineers who build the product are expensive, few in number, and often joining pre-product startups where the whole team is 5-10 people. Once a company has a product and customers, it scales sales, not ML researchers.
If you have a background in B2B enterprise sales, you are already qualified for a large slice of the AI job market. The AI domain knowledge is learnable on the job. The sales motion is not.
The range of real job titles
A sample of actual job titles from the dataset, organized by function:
Sales & Partnerships
- Senior Account Executive
- SDR (Sales Development Representative)
- Revenue Enablement Lead
- Partner Account Executive
Marketing & Growth
- Product Marketing Manager
- Content & Marketing Ops Specialist
- Head of Demand Gen
- Employer Brand Storyteller
Operations
- FP&A Operations Associate
- Director of Operations
- Founding Community Ops Lead
Design
- Senior Product Designer
- Staff Designer — Art Direction
- Design Imagineer
HR & People
- Senior Talent Acquisition Partner
- Technical Recruiter
- Talent Engineering Lead
Finance
- Manager / Director, Finance & Strategy
- GL Accounting Manager
- Finance Deployment Architect
Legal
- Senior Corporate Counsel
- Compliance Engineer
- Associate General Counsel, Commercial & IP
None of these roles requires writing code. Several carry senior-level pay in line with the AI industry premium. Titles like "Finance Deployment Architect" and "Compliance Engineer" show how AI companies rebrand traditional functions, but the underlying work is finance and legal compliance, not engineering.
When non-engineering roles peak
The composition of non-engineering hiring changes a lot across company size, and the pattern tracks growth stage:
| Company Size (Employees) | Non-Engineering % |
|---|---|
| 1-10 | 40.0% |
| 11-50 | 47.1% |
| 51-200 | 46.6% |
| 201-500 | 46.5% |
| 501-1,000 | 56.2% |
| 1,000-5,000 | 47.0% |
| 5,000+ | 37.9% |
The 501-1,000 employee band is the outlier. At this stage, companies have moved past early product development and are aggressively scaling go-to-market: building out sales territories, expanding marketing, adding customer success headcount, and standing up the legal and compliance infrastructure needed to sell to enterprises. That GTM expansion tilts the hiring mix decisively toward non-engineering.
At the very large end (5,000+), the non-engineering share drops to 37.9%. Large AI companies like Anthropic, OpenAI, or Cohere employ proportionally more researchers and engineers as they keep pushing the underlying technology forward. The infrastructure matures too. You need fewer recruiters per employee once HR systems are established.
For job seekers, the 201-1,000 employee range is the sweet spot for non-technical candidates. These are companies with real products, real revenue, and aggressive hiring plans for commercial functions. Junior roles run the other way: our entry-level analysis found that Seed and Series A companies are 3-5x more likely to hire entry-level than late-stage companies. The best stage depends on whether you're switching functions or switching seniority.
By funding stage
Funding stage tells a related but distinct story:
| Funding Stage | Non-Engineering % |
|---|---|
| Bootstrapped | 59.1% |
| Private Equity | 56.4% |
| Series D | 52.2% |
| Pre-Seed | 52.2% |
| Series F | 50.4% |
| Series B | 49.1% |
| Series C | 47.8% |
| Series A | 47.5% |
| Seed | 37.4% |
| Series G+ | 38.4% |
Bootstrapped companies hire the most non-technical staff proportionally (59.1%). The likely reason is that bootstrapped AI companies tend to be services-oriented businesses, consulting, implementation, training, rather than deep-tech product companies. They need operators, sales, and delivery talent more than researchers.
Seed-stage companies are the opposite at 37.4% non-technical. Founders are typically technical, they're building the core product, and the first few hires are usually engineers. The non-technical build-out comes later.
From Seed through Series A, B, and C, the non-technical share climbs steadily as companies add commercial infrastructure alongside continued product development, then plateaus around 48-52% in the Series C to D range.
What this means if you're job hunting
The AI-specific premium is in domain fluency, not technical skills. Sales, marketing, finance, and legal at AI companies work essentially the same as at any SaaS company. What sets candidates apart is being able to talk credibly about what the product does: accuracy versus hallucination, data privacy concerns, enterprise procurement objections. That's learnable in weeks, not years. Your existing functional expertise is the hard part. The AI context is the easy part.
The 201-1,000 employee range is where to aim. These companies hire non-engineering roles at the highest rates (46-56%), with real products, real revenue, and commercial teams under active construction. Seed-stage companies, at 37.4% non-engineering, are still in build mode and skew toward engineers.
Your vertical matters more than "AI experience." If you've sold software to healthcare providers, an AI company targeting that same buyer is a natural fit. If you've done compliance at a fintech, AI companies navigating data regulation want exactly that background. The industry expertise you already have is what AI startups are hiring for. They just happen to build AI products.
If you're early-career, see our companion analysis on where the 1.3% of entry-level AI startup jobs actually are: Only 1.3% of AI Startup Jobs Are Entry-Level.
Methodology
The data underlying this analysis comes from the Fast AI Startup Jobs dataset, which tracks active job listings across AI startups in real time. The dataset included 35,450 job listings across 1,542 AI companies at the time of this analysis.
Jobs were categorized into functional departments using a combination of title parsing and employer-provided category data. The "General / Unclassified" category captures roles that do not map clearly to a standard function. These are real open positions, not data errors.
"Engineering, Research & Product" includes software engineering, ML/AI research, data science, data engineering, DevOps/infrastructure, and product management. All other functions are classified as non-engineering. Product management sits on the boundary of technical and non-technical work; we have included it in the technical bucket, which makes our 46.6% non-engineering figure conservative.
A few limitations to keep in mind. The dataset is sourced from major ATS platforms (Ashby, Greenhouse, Lever, and others). Companies that post jobs only on their own websites, or use ATS platforms we don't track, are not represented. Department classification relies on a combination of title parsing and employer-provided tags, and the 13.5% "General / Unclassified" bucket is a reminder that these labels are imperfect. Some roles in the non-engineering count may involve technical skills (e.g., Sales Engineering, IT), just as some roles in the engineering count may be non-technical in practice.
Company size and funding stage data comes from Fast AI Startup Jobs's company enrichment pipeline. Percentages may not sum exactly to reported totals due to rounding.
Data sourced from the Fast AI Startup Jobs dataset of 35,450 AI startup job listings across 1,542 companies.
Related: We Analyzed 35,000 AI Startup Job Listings. Only 1.3% Are Entry-Level. A deeper look at the 458 entry-level openings: where they are, which funding stages hire juniors, and how to break in.