Which roles AI companies open, which tools and skills their postings ask for, where the work is and what it pays, counted from the open postings on the companies' own public job boards. Companies and roles only, never people. As of Nov 12, 2026. Tracking since Nov 12, 2026.
18 companies4,234 open postings2,010 analysedsince JSON
Open postings
4,234
at 18 companies; 2,010 (47%) analysed so far
Sell and deploy
45%
898 of 2,010 analysed roles sell, deploy or look after customers
Three companies
57%
Databricks, OpenAI and Anthropic hold 2,409 of 4,234 open postings
Top median pay
$425K
Research; 2× customer success, the lowest at $213K
Each company sits in the sector of the role family it hires most for, on the ring of its size. 2,010 of 4,234 open postings are analysed so far; the band shows them by family.
Sell and deploy is the largest analysed group at 13 of 18 companies; only xAI, Perplexity, Cerebras, Thinking Machines Lab and Replit lead with engineering
Bubble area = share of a company's analysed postings in that family. Click a company row or a family column: the title and the three panels below follow the selection.
SelectionCompany page →Rows and column heads are buttons · Esc clears
Sell and deployEngineeringOperationsModels and safetyProduct and design
Companies by open postingsopen · new 30d
Sell and deploy
Engineering
Operations
Models and safety
Product and design
open postings, dark = opened in 30 days
–not in its six largest families
A · Family mixAll 18 companies
Sell and deploy is 45% of analysed roles: from 83% at Databricks to 0% at Thinking Machines Lab
The four largest boards against the industry; each bar is 100% of analysed roles, labels are sector shares.
All companies2,010 analysed45%25%16%
Databricks311 of 913 in its top six83%
OpenAI311 of 823 in its top six52%18%21%
Anthropic249 of 673 in its top six44%26%20%
xAI119 of 299 in its top six67%18%
Company bars use each company's six largest analysed families; the rest of each board is not analysed yet or outside the top six.
B · Tool stackAll 18 companies
Python is asked in 11% of analysed postings, by all 18 companies; Apache Spark comes from one company
Share of analysed postings asking for each tool; right = companies asking, of 18.
Python11% · 18/18
AWS5.0% · 13/18
SQL4.3% · 14/18
Kubernetes4.0% · 16/18
Salesforce3.8% · 12/18
Azure3.4% · 10/18
GCP3.3% · 13/18
Go2.5% · 12/18
Apache Spark2.4% · 1/18
TypeScript2.0% · 10/18
C · Pay bandsAll 18 companies
11 of 15 family medians sit within 15% of the $284K median family; research reaches $615K at its 75th percentile
Bar = 25th to 75th percentile of the USD ranges postings state per year, dot = median.
median family $284K
Research$425K
Data$330K
Hardware$321K
AI engineering$313K
ML engineering$310K
Security$295K
Infrastructure$289K
Software engineering$284K
Safety and policy$280K
Product$280K
Design$270K
Go-to-market$264K
Forward deployed$243K
Operations$229K
Customer success$213K
$200K$300K$400K$500K$600K
Bands need ten postings of one family stating a range (966 in all).
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Company shares use the six largest analysed families the export lists per company (1,653 of 2,010 analysed postings); a dash means "not in its top six", not zero.
Research pays the most, a $425K median on 63 analysed open roles; go-to-market has 8× the openings at $264K
Each bubble is a role family: across = analysed open postings (log scale), up = median stated pay, area = postings opened in the last 30 days, line = 25th to 75th percentile.
← swipe the chart →
Sell and deploy
Engineering
Operations
Models and safety
Product and design
25th–75th percentile
area = opened in 30 days
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Pay: midpoints of the ranges postings state, USD per year, families with ten or more such postings. Bubbles are nudged sideways only where they would touch.
Table: salary bands · 15 rows
Family
Currency
Postings
25th pct
Median
75th pct
Go-to-market
USD per year
182
$207,625
$263,725
$323,000
Operations
USD per year
170
$183,250
$229,000
$265,000
Forward deployed
USD per year
105
$213,750
$242,500
$277,500
Infrastructure
USD per year
85
$229,500
$288,500
$355,500
Safety and policy
USD per year
64
$242,375
$280,000
$330,000
AI engineering
USD per year
60
$263,625
$312,500
$396,000
Product
USD per year
51
$245,500
$280,000
$327,500
Security
USD per year
51
$233,350
$295,000
$362,500
Software engineering
USD per year
41
$226,000
$283,500
$355,500
Customer success
USD per year
32
$172,400
$212,500
$268,375
Data
USD per year
31
$235,500
$330,000
$390,500
Research
USD per year
26
$310,000
$425,000
$615,000
ML engineering
USD per year
23
$288,500
$310,000
$399,250
Design
USD per year
23
$206,750
$270,000
$327,750
Hardware
USD per year
22
$231,875
$320,500
$386,000
Exhibit 3 · Share of analysed roles
Only 10 in 100 analysed roles build the models themselves; 62 sell or support AI products
100 dots = all analysed postings; each dot is one percent. Switch between AI focus, region and work mode above.
CategoryPostingsof 100
Core AI: builds models20610
Applied AI: builds products on models50225
AI-adjacent: sells or supports AI products1,23862
No AI focus643
United States1,24262
UK and Ireland21311
Asia-Pacific19910
Europe1648
Canada693
5 more regions1236
Hybrid55127
On-site35818
Remote33617
Not stated76538
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. "Not stated" is kept as its own share, never re-allocated.
Table · 14 rows
View
Category
Postings
Dots of 100
AI focus
Core AI: builds models
206
10
AI focus
Applied AI: builds products on models
502
25
AI focus
AI-adjacent: sells or supports AI products
1,238
62
AI focus
No AI focus
64
3
Region
United States
1,242
62
Region
UK and Ireland
213
11
Region
Asia-Pacific
199
10
Region
Europe
164
8
Region
Canada
69
3
Region
5 more regions
123
6
Work mode
Hybrid
551
27
Work mode
On-site
358
18
Work mode
Remote
336
17
Work mode
Not stated
765
38
Exhibit 4 · The tools postings ask for
Python is in 216 postings at all 18 companies, 2.1× AWS; only 20 of 60 tools are asked by half the companies or more
Each bubble is a tool: area = postings asking for it, ring = share of companies asking. Hover for the months.
Languages535 mentions · 10 toolsCloud and infra473 mentions · 14 toolsAI and models188 mentions · 9 toolsData platforms180 mentions · 10 toolsSales and workplace166 mentions · 9 toolsWeb and APIs121 mentions · 6 toolsOther tools18 mentions · 2 tools
ring = share of the 18 companies asking · area = postings
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Tool groups are our own grouping by name. 3 tools come from a single company. November counts in the tooltips are month to date.
Table: tools · 60 rows
Tool
Postings
Companies
Share
By month first seen
Python
216
18
11%
Sep 2026 58Oct 2026 108Nov 2026 50
AWS
101
13
5.0%
Sep 2026 30Oct 2026 40Nov 2026 31
SQL
86
14
4.3%
Sep 2026 15Oct 2026 46Nov 2026 25
Kubernetes
81
16
4.0%
Sep 2026 15Oct 2026 48Nov 2026 18
Salesforce
77
12
3.8%
Sep 2026 13Oct 2026 47Nov 2026 17
Azure
68
10
3.4%
Sep 2026 18Oct 2026 29Nov 2026 21
GCP
66
13
3.3%
Sep 2026 21Oct 2026 24Nov 2026 21
Go
50
12
2.5%
Sep 2026 11Oct 2026 27Nov 2026 12
Apache Spark
49
1
2.4%
Sep 2026 12Oct 2026 23Nov 2026 14
TypeScript
41
10
2.0%
Sep 2026 13Oct 2026 16Nov 2026 12
MCP
34
8
1.7%
Sep 2026 8Oct 2026 18Nov 2026 8
APIs
33
12
1.6%
Sep 2026 11Oct 2026 18Nov 2026 4
C++
33
8
1.6%
Sep 2026 8Oct 2026 11Nov 2026 14
Terraform
33
13
1.6%
Sep 2026 5Oct 2026 23Nov 2026 5
Rust
31
10
1.5%
Sep 2026 9Oct 2026 12Nov 2026 10
Claude
29
3
1.4%
Sep 2026 9Oct 2026 16Nov 2026 4
PyTorch
29
9
1.4%
Sep 2026 9Oct 2026 6Nov 2026 14
React
29
10
1.4%
Sep 2026 10Oct 2026 14Nov 2026 5
API
26
10
1.3%
Sep 2026 10Oct 2026 15Nov 2026 1
Java
25
5
1.2%
Sep 2026 4Oct 2026 14Nov 2026 7
Linux
25
11
1.2%
Sep 2026 6Oct 2026 12Nov 2026 7
JavaScript
23
7
1.1%
Sep 2026 9Oct 2026 10Nov 2026 4
JAX
22
6
1.1%
Sep 2026 8Oct 2026 5Nov 2026 9
LLM
21
9
1.0%
Sep 2026 4Oct 2026 12Nov 2026 5
PostgreSQL
21
7
1.0%
Sep 2026 4Oct 2026 13Nov 2026 4
Slack
21
11
1.0%
Sep 2026 1Oct 2026 12Nov 2026 8
Databricks
18
4
0.9%
Sep 2026 3Oct 2026 11Nov 2026 4
Scala
18
2
0.9%
Sep 2026 4Oct 2026 9Nov 2026 5
CI/CD
17
12
0.8%
Sep 2026 5Oct 2026 8Nov 2026 4
Delta Lake
17
1
0.8%
Sep 2026 5Oct 2026 7Nov 2026 5
LLMs
17
8
0.8%
Sep 2026 3Oct 2026 10Nov 2026 4
GPU
16
8
0.8%
Sep 2026 5Oct 2026 7Nov 2026 4
Next.js
15
6
0.7%
Sep 2026 5Oct 2026 6Nov 2026 4
BigQuery
14
6
0.7%
Sep 2026 2Oct 2026 8Nov 2026 4
Kafka
14
5
0.7%
Sep 2026 3Oct 2026 4Nov 2026 7
Spark
14
4
0.7%
Sep 2026 3Oct 2026 3Nov 2026 8
Claude Code
13
5
0.6%
Sep 2026 5Oct 2026 6Nov 2026 2
GitHub
13
9
0.6%
Sep 2026 1Oct 2026 8Nov 2026 4
Google Cloud
13
5
0.6%
Sep 2026 2Oct 2026 5Nov 2026 6
Google Workspace
13
8
0.6%
Oct 2026 12Nov 2026 1
dbt
13
6
0.6%
Sep 2026 1Oct 2026 9Nov 2026 3
Bash
12
7
0.6%
Oct 2026 10Nov 2026 2
Codex
12
2
0.6%
Sep 2026 7Oct 2026 3Nov 2026 2
Flink
11
4
0.5%
Sep 2026 2Oct 2026 1Nov 2026 8
InfiniBand
11
3
0.5%
Sep 2026 1Oct 2026 9Nov 2026 1
LinkedIn
11
6
0.5%
Sep 2026 2Oct 2026 3Nov 2026 6
MLflow
11
1
0.5%
Sep 2026 1Oct 2026 6Nov 2026 4
SCADA
11
3
0.5%
Sep 2026 5Oct 2026 6
Helm
10
2
0.5%
Sep 2026 1Oct 2026 9
Android
9
4
0.4%
Sep 2026 1Oct 2026 5Nov 2026 3
Ashby
9
7
0.4%
Oct 2026 6Nov 2026 3
Ethernet
9
4
0.4%
Sep 2026 3Oct 2026 6
Git
9
5
0.4%
Sep 2026 2Oct 2026 4Nov 2026 3
Hadoop
9
3
0.4%
Sep 2026 2Oct 2026 3Nov 2026 4
Jira
9
3
0.4%
Sep 2026 1Oct 2026 6Nov 2026 2
Node.js
9
4
0.4%
Sep 2026 3Oct 2026 2Nov 2026 4
Okta
9
6
0.4%
Oct 2026 5Nov 2026 4
SAML
9
6
0.4%
Sep 2026 1Oct 2026 6Nov 2026 2
FedRAMP
8
4
0.4%
Sep 2026 4Oct 2026 4
GKE
8
2
0.4%
Oct 2026 8
Exhibit 5 · Tools on the move
Salesforce postings rose 3.6× from September to October (13 → 47); of 20 tools with 8+ September postings, only PyTorch and JAX fell
Postings still open, by the month they were published; log scale, so equal slopes mean equal growth. November is month to date and drawn apart.
Rose most (top 8 by ratio)
Fell
Nov 1–12: month to date, not comparable
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Tracking began Nov 12, 2026: September roles that closed before then are not counted, so every rise is partly survivorship.
Table · 20 rows
Tool
Sep
Oct
Oct ÷ Sep
Nov to date
Salesforce
13
47
3.6×
17
Kubernetes
15
48
3.2×
18
SQL
15
46
3.1×
25
Go
11
27
2.5×
12
MCP
8
18
2.3×
8
Apache Spark
12
23
1.9×
14
Python
58
108
1.9×
50
Claude
9
16
1.8×
4
APIs
11
18
1.6×
4
Azure
18
29
1.6×
21
API
10
15
1.5×
1
React
10
14
1.4×
5
C++
8
11
1.4×
14
AWS
30
40
1.3×
31
Rust
9
12
1.3×
10
TypeScript
13
16
1.2×
12
GCP
21
24
1.1×
21
JavaScript
9
10
1.1×
4
PyTorch
9
6
0.7×
14
JAX
8
5
0.6×
9
Exhibit 6 · Job titles
Software Engineer is the single biggest title (166 postings at 17 of 18 companies), but sell and deploy titles fill 50% of the top 40
Area = open postings with that exact title, without level, team or location; colour = sector read from the title.
Sell and deploy
Engineering
Operations
Models and safety
Product and design
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Colour is inferred from title keywords, not from the family classifier. 910 postings across the 40 titles. The table links example postings on the company's own board.
Enterprise sales leads with 243 postings at 16 of 18 companies; observability is the most widespread, asked at 17
Bar = postings asking for the skill; dots = companies asking, of 18. Top 20 of 60.
companies asking, of 18
enterprise sales24316
evals14816
stakeholder management12713
solution architecture11711
distributed systems10716
pipeline generation9013
pipeline management8312
observability7817
vendor management7615
account management6710
program management6711
data pipelines6613
incident response6614
solutions architecture666
data engineering635
capacity planning6111
team leadership5816
forecasting5611
contract negotiation5311
machine learning515
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. What the work itself asks for, counted once per posting.
Table: skills · 60 rows
Skill
Postings
Companies
Share
enterprise sales
243
16
12%
evals
148
16
7.4%
stakeholder management
127
13
6.3%
solution architecture
117
11
5.8%
distributed systems
107
16
5.3%
pipeline generation
90
13
4.5%
pipeline management
83
12
4.1%
observability
78
17
3.9%
vendor management
76
15
3.8%
account management
67
10
3.3%
program management
67
11
3.3%
data pipelines
66
13
3.3%
incident response
66
14
3.3%
solutions architecture
66
6
3.3%
data engineering
63
5
3.1%
capacity planning
61
11
3.0%
team leadership
58
16
2.9%
forecasting
56
11
2.8%
contract negotiation
53
11
2.6%
machine learning
51
5
2.5%
full-stack development
50
11
2.5%
customer delivery
45
7
2.2%
go-to-market strategy
44
12
2.2%
partner enablement
42
10
2.1%
reinforcement learning
42
10
2.1%
people management
41
11
2.0%
outbound prospecting
40
10
2.0%
customer adoption
39
7
1.9%
prototyping
39
5
1.9%
risk management
39
10
1.9%
commissioning
38
4
1.9%
solution selling
38
3
1.9%
customer discovery
37
7
1.8%
product strategy
37
9
1.8%
technical pre-sales
37
4
1.8%
workflow automation
37
12
1.8%
account planning
36
8
1.8%
automation
35
7
1.7%
executive relationships
35
8
1.7%
account strategy
34
6
1.7%
experimentation
34
9
1.7%
saas sales
34
8
1.7%
ai agents
33
13
1.6%
cloud infrastructure
32
12
1.6%
consultative selling
32
7
1.6%
customer engineering
32
8
1.6%
customer success
32
10
1.6%
data governance
32
9
1.6%
pre-sales
32
7
1.6%
data quality
31
12
1.5%
proofs of concept
31
6
1.5%
technical discovery
31
6
1.5%
cloud sales
30
1
1.5%
data center operations
30
4
1.5%
sales enablement
30
11
1.5%
ci/cd
29
9
1.4%
customer engagement
29
5
1.4%
post-training
29
11
1.4%
technical leadership
29
11
1.4%
ai safety
28
4
1.4%
Exhibit 8 · Hiring freshness
Cursor opened 91% of its 113 open roles in the last 30 days; no other company tops 52%, and at Databricks, the largest board, it is 21%
Position = share of each company's open postings published in the last 30 days; area = open postings; colour = its largest analysed sector. Click a bubble for the company page.
← swipe the chart →
Sell and deploy
Engineering
No clear lead or not analysed
All 18 companies: 28%
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. Open on our first day without a board date counts as baseline, not new.
Senior is 49% of postings that state a level; for every junior or intern opening there are 9.3 senior ones
Column height = analysed open postings at each level: individual-contributor rungs, then management; postings that state no level stand apart.
21InternInt1.7%
45JuniorJr3.6%
137MidMid11%
614SeniorSr49%
107Staff+Staff+8.5%
Individual contributor
230ManagerMgr18%
106Director+Dir+8.4%
Management
750Not statedNonen/a
Set apart
Source: the public job boards of 18 AI companies (238 boards), read daily · as of Nov 12, 2026. 37% of analysed postings state no level; they are never re-allocated.
Table · 8 rows
Level
Postings
Share of stated
Intern
21
1.7%
Junior
45
3.6%
Mid
137
11%
Senior
614
49%
Staff+
107
8.5%
Manager
230
18%
Director+
106
8.4%
Not stated
750
—
Exhibit 10 · Company workbench · open a company
Go-to-market is the largest analysed family at 11 of 18 boards; the analysed top six cover from 25% of Scale AI to 84% of Cursor so far
Ring = all open postings of the company; coloured arcs = its six largest analysed families; the dark rest is not analysed yet or smaller. Each card opens the company page.
Counts come from the public job boards of AI companies (Greenhouse, Ashby and Lever), read once a day.
A posting is open from the day we first see it until a full read of its board no longer lists it.
Each posting is classified by a language model into a fixed set of role families, seniority levels, work modes and AI focus, and the tools and skills it asks for are listed; shares are of classified open postings.
New means published in the last 30 days (or first seen after our first collection day when the board gives no date); everything open on our first day is the baseline, not new hiring.
Salary bands use the midpoint of ranges the postings state themselves, only where at least 10 postings of one family state a range in one currency.
Roles, never people: no recruiter, hiring manager or applicant is collected.
Who read this page?Machine requests, hidden until opened+