Skills demand · updated 2026-09-28

AI skills in demand 2026: what 286 AI companies ask for

Tools, frameworks and skills named in 14,904 open job descriptions at AI labs, startups and infrastructure companies, refreshed every 6 hours. Share is the fraction of descriptions that mention the skill.

Most requested in 14,904 job descriptions

Python4,174

LLMs2,247

Agents2,059

AWS1,635

Robotics1,562

Kubernetes1,506

C++1,369

GCP1,314

Azure1,185

SQL1,153

TypeScript1,050

Simulation1,025

PyTorch713

React677

Terraform630

Security clearance613

Salesforce597

Rust563

Evals563

Spark536

All tracked skills

Skill	Roles mentioning it	Share of descriptions
Python	4,174	28.0%
LLMs	2,247	15.1%
Agents	2,059	13.8%
AWS	1,635	11.0%
Robotics	1,562	10.5%
Kubernetes	1,506	10.1%
C++	1,369	9.2%
GCP	1,314	8.8%
Azure	1,185	8.0%
SQL	1,153	7.7%
TypeScript	1,050	7.0%
Simulation	1,025	6.9%
PyTorch	713	4.8%
React	677	4.5%
Terraform	630	4.2%
Security clearance	613	4.1%
Salesforce	597	4.0%
Rust	563	3.8%
Evals	563	3.8%
Spark	536	3.6%
Docker	504	3.4%
Fine-tuning	493	3.3%
Postgres	374	2.5%
Reinforcement learning	366	2.5%
RAG	356	2.4%
Transformers	331	2.2%
Grafana	311	2.1%
Computer vision	292	2.0%
CUDA	285	1.9%
MLOps	285	1.9%
Distributed training	284	1.9%
TensorFlow	271	1.8%
Prometheus	255	1.7%
SOC 2	252	1.7%
JAX	249	1.7%
HubSpot	218	1.5%
Kafka	216	1.4%
Snowflake	209	1.4%
InfiniBand	196	1.3%
PCB	193	1.3%

One page per skill: companies asking for it and the newest roles.

By function

Engineering

Skill	Roles	Share
Python	2009	46.2%
LLMs	1014	23.3%
Agents	891	20.5%
C++	815	18.7%
TypeScript	791	18.2%
AWS	767	17.6%
Kubernetes	688	15.8%
Robotics	635	14.6%

Product & Design

Skill	Roles	Share
Agents	170	23.6%
Figma	110	15.3%
LLMs	105	14.6%
Robotics	52	7.2%
TypeScript	50	6.9%
React	41	5.7%
SQL	38	5.3%
Evals	38	5.3%

Infrastructure

Skill	Roles	Share
Python	994	40.6%
Kubernetes	541	22.1%
C++	413	16.9%
AWS	376	15.4%
Robotics	372	15.2%
Terraform	314	12.8%
GCP	298	12.2%
Simulation	289	11.8%

Go-to-market

Skill	Roles	Share
Salesforce	430	12.9%
Agents	406	12.2%
Python	379	11.4%
LLMs	372	11.2%
AWS	320	9.6%
SQL	310	9.3%
Azure	279	8.4%
GCP	271	8.1%

Operations

Skill	Roles	Share
Robotics	226	9.3%
SQL	145	5.9%
LLMs	119	4.9%
Agents	113	4.6%
Python	110	4.5%
Security clearance	72	2.9%
Salesforce	45	1.8%
SOC 2	43	1.8%

Research

Skill	Roles	Share
Python	413	43.0%
LLMs	318	33.1%
PyTorch	261	27.2%
Agents	219	22.8%
Reinforcement learning	169	17.6%
Fine-tuning	147	15.3%
JAX	135	14.0%
Robotics	133	13.8%

Data

Skill	Roles	Share
Python	201	55.2%
SQL	188	51.6%
LLMs	74	20.3%
dbt	69	19.0%
Snowflake	44	12.1%
Robotics	42	11.5%
Spark	41	11.3%
Agents	39	10.7%

Skills by company (which of the 252 companies ask for CUDA, JAX, Kubernetes or a clearance, and how often), as a table and a CSV, are in AI Hiring Index Pro, $149/month. Vendors use it to see where their category is being hired for; candidates use it to see which company stack matches theirs.

How it is computed

Each description is matched against a fixed list of 69 skill and tool names with word-boundary patterns (so "Go" the language is not "go" the verb). A role counts once per skill however many times it mentions it. Descriptions are the companies' own text from their careers systems, truncated at 8,000 characters.