Skip to main content

AI Hiring Index · as of 2026-11-11

9,362 open roles at 80 AI companies

What AI companies actually hire for — measured every month from their own job posts.

For software engineers with five or more years who are weighing a move into an AI lab or an AI-native company: which roles are open, what each one asks for, and what you would still have to learn.

Technical roles5,059 · 54%

Everything else4,303 · 46%

  • GTM1,829
  • G&A1,146
  • Program / Ops451
  • Product / Design328
  • Field ops / Manufacturing190
  • Other359

858 new in the last 7 days · browse the open roles · the November 2026 Index, with month-over-month change

Who is hiring

The 80 companies we track, by kind. Each company’s page shows which families it hires for and how much of its hiring is technical.

Kind of companyCompaniesOpen rolesHiring most
AI infrastructure203,116Databricks 908, Nebius 373, Crusoe 323
AI lab162,487OpenAI 829, Anthropic 679, xAI 301
AI-native app242,356Harvey 373, Legora 327, Sierra 213
Robotics101,129Applied Intuition 320, Wayve 166, Field AI 102
AI developer tools10274LangChain 100, Exa 66, Braintrust 32

All 80 companies, and when each board was last read

Where an experienced engineer fits

The roles this site is built for. Each one’s page has the full numbers, the open postings and a route through the learning map.

Forward Deployed Engineer

Forward deployed and applied AI engineers take a customer's real workflow and build the LLM application that runs it on their company's platform — agents, retrieval, evals — through to production, then carry what they learned back to the product team.

Open roles
1,095
Posted base
$182K–$250K · 44% post pay
Remote
27%
Hiring most
Databricks, OpenAI, Anthropic

What you can skip

Shipping production software, integrating APIs, debugging a system you did not write, cloud and containers, and talking to customers about trade-offs — the baseline half of these postings is what a senior software engineer already does.

AI Engineer

AI engineers build the product features that run on language models and the systems underneath them — retrieval, agents, evaluation, and fine-tuning where prompting is not enough. At model companies many also train and serve models, which is why PyTorch appears in so many of these postings.

Open roles
176
Posted base
$196K–$325K · 55% post pay
Remote
10%
Hiring most
Sierra, Wayve, Nebius

What you can skip

Backend and API design, data modelling, testing, deployment and on-call — the software engineering these roles are built on. What changes is that model behaviour is now part of the system you test.

Software Engineer

Software engineers at AI companies build the same things as anywhere — product, platform, infrastructure — around a model at the centre: serving it, putting it in front of users, and building the tools researchers train it with.

Open roles
1,502
Posted base
$190K–$275K · 59% post pay
Remote
9%
Hiring most
Databricks, OpenAI, Applied Intuition

What you can skip

All of it is your existing job. The difference is context: expect to work next to model behaviour (latency, cost, non-determinism) and to use AI tools in your own work.

What each role asks for

How many of each family’s postings name a skill, in percent. Read down a column for the shape of a role: the first block is what a move into AI adds, the second is engineering most senior engineers already do.

Target rolesOther technical families
SkillAll technicalFDE / AppliedAI / ML EngineerSoftware EngInfra / HardwareSecurityResearch EngineerResearch ScientistInference / PerfEvals / DataSafety / Policy
Open roles5,0591,0951761,5021,112318244182179138113
Median posted base—$216K$284K$232K$245K$265K$300K$300K$250K$270K$266K
The AI part — what the learning map teaches
LLMs25%39%55%16%7%19%58%46%42%31%19%
Agents20%32%45%17%9%19%35%30%4%16%10%
Evals20%28%60%12%5%4%60%62%15%37%19%
PyTorch8%4%26%3%3%<1%36%38%31%4%2%
RL / post-training7%3%18%3%1%<1%53%43%6%6%3%
RAG / retrieval7%16%35%5%2%1%6%2%0%2%0%
Fine-tuning5%7%19%2%1%0%23%18%3%4%2%
Prompting4%11%24%3%<1%<1%2%0%0%4%7%
Distributed training3%1%10%1%2%1%23%16%6%1%2%
CUDA / Triton3%1%7%1%2%0%9%6%31%0%0%
vLLM / SGLang / TensorRT3%3%6%1%1%<1%6%5%27%0%0%
JAX3%<1%5%1%1%0%17%17%9%1%2%
MCP2%3%5%2%1%2%2%0%0%0%0%
General engineering — most senior engineers have it
Python48%56%52%45%40%49%50%58%64%60%26%
Kubernetes19%11%15%18%26%35%17%4%30%1%3%
Go14%3%15%19%20%30%8%2%23%7%4%
C++14%3%19%25%12%10%9%8%42%4%4%
TypeScript11%15%24%15%4%10%5%1%3%8%9%
Rust8%1%9%10%9%13%9%2%21%4%5%
Credential
PhD5%3%9%3%3%3%9%44%6%7%4%
Dot area is the share:5%25%50%Median posted base: the middle of each posting’s published range, shown where at least 10 postings publish one.

Learn the part that’s new

The learning map teaches the AI block above, in the order you need it: the foundations every route shares, then a route for your role. Each stop has a few resources, picked by hand.

Foundations

  1. How the models work
  2. Call a model from code
  3. Search by meaning
  4. Machine learning, the classical way
  5. Measure it

Then the route for your role:

Forward Deployed Engineer

  1. Build on the customer's data and tools
  2. Prove it works
  3. Deploy into their environment
  4. Adapt the model

AI Engineer

  1. Build LLM features
  2. Measure and improve
  3. Train and post-train
  4. Serve it

Open the learning map — how a language model works, the knowledge around it in layers, and every route in full.

How these numbers are made

Measured, not estimated. Each rule links to its section of the methodology, which has the rest.

From each company’s own job board
Every night we read the public Ashby, Greenhouse and Lever boards of the 80 companies we track; 80 answered on 2026-11-11. These are posted openings, not hires.
Facts, never the job description
For each posting we keep its title, location, posted pay, role family and skills, and a link to the company’s own posting. The description itself is not stored or shown.
Posted base pay, not total compensation
A pay figure appears only where at least 10 postings publish a range; 53% of the postings in the roles above do. Equity and bonus are in none of them.
Skills are audited every month
Skills come from a fixed list of 20. Every month a random sample of postings is checked tag by tag against the original, and a skill whose tags are right less than 90% of the time is not shown.
A failed board changes nothing
A board that does not load closes none of its postings, and drops out of every number after 3 days. Month-over-month changes compare only companies measured in both months.

The AI Hiring Index by email

One email a month, when an issue is published — the numbers and a link to the full issue. Double opt-in, and every email has a one-click unsubscribe (privacy).

Spot a mistake on this page?