Hiring data as of October 28, 2026

The NYC AI skill mapwho asks for what

Name: NYC AI Skill Map
Creator: Neuronify

Which AI skills New York’s fintech, healthcare, infrastructure, media and enterprise software companies hire for, which are rising, which have spread beyond engineering, and what the roles pay. Read from 46 companies’ job boards every morning.

See the map

NYC AI companies read: 46NYC AI companies read
open roles read today: 1,559open roles read today
skills and tools tracked: 81skills and tools tracked
sector groups compared: 6sector groups compared

Why a map, not a list

The AI skills glossary ranks every term by how many New York AI companies name it. That answers what the market as a whole asks for. It hides a second question that matters just as much to a candidate choosing where to apply, or a company deciding what to hire for: does a healthcare AI company ask for the same skills as a fintech or a voice-AI start-up?

The map answers it by splitting the same daily data by the kind of company asking. Rows are skills and tools; columns are sector groups; each cell is the number of companies in that group naming the term in at least one open role. The shade shows the share of the group, so a dark cell means most of that sector asks for it. Python is named by 40 of 46 companies, more than any other term.

Three cuts follow the map: the skills each sector names far more than the market does, the skills whose share of companies has moved most since daily company counts began, and the AI terms that now appear in sales, product and operations roles as often as in engineering ones. The last section shows posted pay for roles naming each term.

The map: 30 most-named skills by sector

Companies naming each term in at least one open role, by sector group. Column headers show how many companies each group has. Cells are shaded by share of the group; groups with fewer than 3 companies are shown but should be read as anecdotes.

AI skills by sector group, number of companies naming each
Skill or tool	All46 cos.	Fintech & insurance11 cos.	Healthcare9 cos.	AI infrastructure7 cos.	Media & voice5 cos.	Enterprise software11 cos.	Other3 cos.
PythonLanguages	40	11	7	6	4	10	2
Machine learningModel development	39	11	6	7	4	11	·
AI agents (agentic AI)Building with LLMs	38	10	8	7	3	10	·
LLMs (large language models)Building with LLMs	37	11	7	6	4	9	·
AWSCompute	36	8	8	5	3	11	1
TypeScriptLanguages	36	11	8	4	4	8	1
SQLLanguages	33	11	6	4	3	8	1
ObservabilityInfrastructure & operations	31	9	7	6	2	7	·
GoLanguages	29	7	7	4	3	7	1
Evals (LLM evaluation)Building with LLMs	28	6	7	6	3	6	·
Data pipelines (ETL)Infrastructure & operations	27	7	6	3	4	7	·
KubernetesCompute	25	7	7	4	3	4	·
Anthropic (Claude)Model providers	22	4	4	3	3	8	·
GCPCompute	21	4	4	5	3	5	·
Generative AIBuilding with LLMs	20	5	1	4	3	7	·
dbtData platform	19	5	4	1	2	6	1
SnowflakeData platform	18	5	4	2	·	7	·
RAG (retrieval-augmented generation)Building with LLMs	17	3	2	3	2	7	·
AzureCompute	16	2	3	5	2	4	·
InferenceInfrastructure & operations	14	2	1	3	4	4	·
JavaLanguages	14	3	3	3	1	4	·
OpenAI (GPT)Model providers	14	2	4	2	1	5	·
Prompt engineeringBuilding with LLMs	14	4	3	1	2	4	·
MLOps (ML platform)Infrastructure & operations	12	1	2	5	3	1	·
PyTorchTraining frameworks	12	2	·	5	3	2	·
RustLanguages	12	2	1	3	4	2	·
KotlinLanguages	11	3	2	·	2	3	1
MCP (Model Context Protocol)Orchestration & agents	11	4	1	1	2	3	·
Tool use (function calling)Building with LLMs	11	2	·	2	3	4	·
Apache AirflowData platform	10	1	3	2	·	4	·

Source: Neuronify daily read of public job boards, October 28, 2026. Every term links to its glossary entry; the full list of 81 terms is on the glossary page.

Each sector’s signature skills

The core of AI hiring is shared: 6 of the 81 terms we track are named by at least one company in every sector group with three or more companies. What distinguishes a sector is what it names more often than the market does. For each group, these are the terms with the highest share of the group’s companies relative to the share across all 46, among terms at least three of its companies name.

Fintech & insurance · 11 companies

Human-in-the-loop: 4 of 11 companies (36%), against 17% of all companies
AI governance: 3 of 11 companies (27%), against 15% of all companies
MCP (Model Context Protocol): 4 of 11 companies (36%), against 24% of all companies

Healthcare · 9 companies

Google Gemini (Vertex AI): 3 of 9 companies (33%), against 13% of all companies
Apache Airflow: 3 of 9 companies (33%), against 22% of all companies
OpenAI (GPT): 4 of 9 companies (44%), against 30% of all companies

AI infrastructure · 7 companies

Reinforcement learning (RLHF): 3 of 7 companies (43%), against 9% of all companies
Synthetic data: 3 of 7 companies (43%), against 11% of all companies
Multimodal AI: 3 of 7 companies (43%), against 15% of all companies

Media & voice · 5 companies

Distributed training: 3 of 5 companies (60%), against 9% of all companies
CUDA: 3 of 5 companies (60%), against 11% of all companies
GPUs: 3 of 5 companies (60%), against 13% of all companies

Enterprise software · 11 companies

Ruby: 3 of 11 companies (27%), against 9% of all companies
NLP (natural language processing): 3 of 11 companies (27%), against 11% of all companies
Elasticsearch (OpenSearch): 3 of 11 companies (27%), against 15% of all companies

Other · 3 companies

No term is named by three or more of this group’s companies today.

Signature skills say more about what a sector builds than its marketing does. A group whose signature terms are evaluation and governance is putting models in front of regulators and customers; one whose signature terms are inference and GPUs is building the plumbing. For a candidate, the signature list is the shortest route to standing out in a sector: the core skills get you considered everywhere, the signature skills get you considered first.

Rising and falling

Share of scanned companies naming each term, averaged over the first seven days of daily company counts (September 20, 2026 onward) and the latest seven (to October 28, 2026). Shares rather than counts, because the roster itself moves by a few companies. Only terms named by at least five companies at one end or the other are included. The history covers 39 days, so these are early signals: a single company opening a team can move a term several points.

Rising

Evals (LLM evaluation)53% → 61%
Rust19% → 25%
Synthetic data4% → 11%
SQL65% → 71%
Snowflake34% → 39%
Kotlin19% → 24%

Falling

Observability74% → 67%
OpenAI (GPT)39% → 32%
Data pipelines (ETL)64% → 56%
RAG (retrieval-augmented generation)42% → 35%
Google Gemini (Vertex AI)19% → 13%
TypeScript85% → 79%

AI skills beyond engineering

A term in a job post is not only a requirement for engineers. For every term named in at least 40 postings, the table shows the share of those postings that sit outside engineering and research, and the function that names it most among the rest. Generative AI leads: 76% of the postings naming it are outside engineering. AI agents is named by 38 companies.

Share of postings naming each term that are outside engineering
Term	Postings naming it	Outside engineering	Most common other function
Generative AI	79	76%	GTM (41)
AI agents (agentic AI)	247	50%	GTM (58)
SQL	176	45%	GTM (32)
LLMs (large language models)	241	43%	GTM (52)
Anthropic (Claude)	59	37%	GTM (10)
Evals (LLM evaluation)	42	29%	Product (4)
Machine learning	193	25%	GTM (19)
dbt	41	20%	Operations (2)
Data pipelines (ETL)	85	16%	GTM (7)
Python	289	16%	Operations (15)

This is where the AI job market is changing fastest, and where the job titles lag behind. Generative AI appears in 41 GTM postings; AI agents (agentic AI) appears in 58 GTM postings; SQL appears in 32 GTM postings. For people outside engineering, these terms are the practical definition of “AI literacy” in New York hiring: not building models, but working fluently with the products built on them. For companies, a posting outside engineering that names an AI term is a sign the AI product has reached the customer.

What the skills pay

Median posted base salary on New York roles naming each term, from the ranges employers publish under the city’s pay transparency law. Only terms with at least eight distinct posted ranges are shown. For comparison, the median across all New York engineering roles we read is $215k. A term’s median describes the roles that name it, which mixes seniority and function; it is not a premium for the skill itself.

Highest medians

Hugging Face$255k ($158k–$282k, n=8)
Java$250k ($215k–$282k, n=23)
Rust$241k ($215k–$260k, n=12)
Inference$238k ($215k–$275k, n=17)
RAG (retrieval-augmented generation)$238k ($178k–$255k, n=17)
Apache Airflow$235k ($205k–$270k, n=13)
Apache Kafka$231k ($205k–$255k, n=13)
Azure$230k ($200k–$250k, n=21)

Lowest medians

SQL$185k (n=94)
Prompt engineering$180k (n=9)
OpenAI (GPT)$172k (n=18)

Low medians usually reflect where a term is named (sales and operations roles) rather than what the skill is worth. See the NYC AI engineer salary guide for pay by role.

How to use it

If you are looking for a role. Start with the core skills every sector names, then pick a sector and learn its signature terms. The rising list shows where demand is moving before job titles catch up; the beyond-engineering table shows which AI terms you will be asked about even if you never write code.

If you are hiring. Compare your job descriptions with your sector’s column. Naming a tool every peer names doesn’t differentiate your posting; naming one nobody in your sector names may mean you are asking for something the local market doesn’t have, which is worth knowing before a search starts. The pay ladder is a rough check on whether a posted range is competitive for roles that name the same terms.

If you are building or investing. Job posts are a leading indicator of what companies are putting into production. A tool spreading across sectors is gaining adoption; one confined to a single sector is a niche; one falling from the rising list may be losing to an alternative. The data won’t tell you why, but it tells you where to look.

How the map is made

Every morning we read the open roles on the public job boards of the New York AI companies we cover, through the Greenhouse, Ashby and Lever posting APIs. Tools and concepts are extracted from each posting’s own text, with sentences a company repeats across its postings removed so boilerplate doesn’t count, and the text is then discarded. Each company counts once per term. Sector groups fold the NYC AI 100’s thirteen sectors into six. The coverage is weighted toward venture-backed AI companies, so terms that matter most inside banks, insurers and law firms are understated. The full method and a dated change log are on the methodology page.

NYC AI skill map — common questions

Which AI skills are New York companies hiring for?

On October 28, 2026, the most widely named across the 46 New York AI companies we read were Python (40 companies), Machine learning (39 companies), AI agents (agentic AI) (38 companies), LLMs (large language models) (37 companies), AWS (36 companies). The full ranking, with a definition of each term, is in the AI skills glossary.

Do different industries ask for different AI skills?

Yes, at the margin. A core set is named across every sector we can measure, led by Python, AWS, TypeScript. Each sector also has skills it names far more often than the market as a whole; the map’s signature-skills section lists them.

Which AI skills are growing fastest in New York?

Between September 20, 2026 and October 28, 2026, the terms whose share of companies grew most were Evals (LLM evaluation) (53% to 61% of companies), Rust (19% to 25% of companies), Synthetic data (4% to 11% of companies). The window is short, so treat these as early signals rather than trends.

Are AI skills only for engineers?

No. Of the postings naming Generative AI, 76% are outside engineering and research, most of them in GTM. Several AI terms now appear regularly in sales, product, operations and finance roles.

How is a sector assigned?

Each company on the NYC AI 100 has one primary sector. We fold the thirteen sectors into six groups so every column has enough companies to read, and companies from the AI in NYC Show roster are placed by their stated focus. A company counts once per term, however many of its postings name it.

How often is the skill map updated?

Daily. The job boards are read each morning, tags are extracted the same day and the text is discarded, and the page is rebuilt from the new snapshot.

Need these skills on your team?

A skill in a job post is easy to name and hard to verify. We match New York companies with AI engineers and consultancies vetted on production work.

Start a project brief Join the expert network