Skip to main content

Role · AI Hiring Index · as of 2026-11-20

Software Engineer jobs at AI companies

OpenAI (164) and Databricks (159) lead 1,487 open Software Engineer roles across 75 AI companies. 9% are remote and 58% are in the Bay Area. The November 2026 Index counted 4 fewer openings than October 2026, among companies measured in both months.

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.

Every engineering organisation: product, platform and infrastructure, developer productivity, research tooling.

Demand this month

Open roles
1,487
−4
New in the last 7 days
78
Remote
9%
Bay Area
58%

What Software Engineer postings ask for

Share of 1,487 open postings that mention each skill; change since the previous Index in points.

SkillShare, as a barShare
Python44%n=661−1 pt
C++25%n=370+5 pt
Go20%n=294±0 pt
Kubernetes18%n=272−2 pt
Agents16%n=245−2 pt
LLMs16%n=235±0 pt
TypeScript15%n=221−3 pt
Evals12%n=179+4 pt
Rust11%n=161+1 pt
RAG / retrieval5%n=71±0 pt
RL / post-training4%n=54±0 pt
PhD3%n=48+2 pt
PyTorch3%n=44+1 pt
Prompting3%n=40−1 pt
MCP2%n=35+1 pt
Fine-tuning2%n=27+1 pt
CUDA / Triton1%n=21−1 pt
JAX1%n=20+1 pt
Distributed training1%n=20−1 pt
vLLM / SGLang / TensorRT1%n=10±0 pt

Posted base pay

Median posted range $191K–$280K base, from 892 postings that publish pay (60% of this role's openings). This is the base range a posting advertises — not total compensation, which adds equity and bonus; for that, Levels.fyi is the reference.

Midpoint of each posted base range, in thousands of US dollars: <150: 52; 150–200: 198; 200–250: 233; 250–300: 156; 300–350: 92; 350–400: 98; 400–450: 50; 450+: 13

Half of the posted ranges have their midpoint between $195K and $308K; the median midpoint is $238K.

What to learn

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.

The gaps, in the order these postings ask for them

Kubernetes · 18% of postings

Project: Deploy your agent where a customer would run it — Package the agent as a container, deploy it to a local cluster with its secrets, a health probe and resource limits, then kill a node and watch what happens.

Agents · 16% of postings

Project: Write the agent loop yourself, then evaluate it — Messages API tool use with a step cap, tool-error retries and context trimming; score it on 30 real tasks for tool choice, argument correctness and steps taken.

LLMs · 16% of postings

  • Deep Dive into LLMs like ChatGPT — Andrej Karpathy free

    Three and a half hours from tokens to RLHF, from one of OpenAI's founding members — the mental model every later choice rests on.

  • AI Engineering — Chip Huyen · O'Reilly paid

    Chapter 2 is this skill — training data, architecture, post-training, sampling — for people who build on models; the rest is the AI Engineer track at length: evals, RAG, agents, cost.

Project: Ship one LLM feature end to end — Structured output with schema validation, streaming, a fallback model, and per-request cost and latency logged — then write down the numbers.

TypeScript · 15% of postings

  • The TypeScript Handbook free

    The language's own reference, short enough to read in an evening — enough to read and change the web and SDK code these roles ship to customers.

Project: Type an SDK for an API you use — Write a small typed client for a real API — request and response types, errors as values, one streaming endpoint — and publish it with its tests.

One project

Put a model behind a real interface — Build a small service that calls a model with streaming, retries, rate limits and cost tracking, and a UI on top — the plumbing every team here owns some of.

The full learning map

Open Software Engineer roles

See all 1,501 postings

Posted demand, not hires: open postings on the public job boards of the AI companies we track, read every night. A repost counts once; a role family is assigned from the title; skill tags are checked against the postings every month and a skill below 90% precision is not shown; pay is the posted base range, shown only when at least 10 postings publish it. How we count.

Spot a mistake on this page?