San Francisco AI Jobs in September 2026
Strip OpenAI from San Francisco’s AI job market and the city drops below New York. 163 open AI/ML roles make SF the second-largest US market, but 75 of them — 46% — belong to a single employer. Data is from the September 21, 2026 HireIndex snapshot.
How many AI jobs are open in San Francisco right now?
163, across 40 companies. That puts San Francisco second in the US and third globally:
| City | Open AI/ML roles |
|---|---|
| London | 206 |
| San Francisco | 163 |
| Toronto | 121 |
| New York | 106 |
| Sydney | 71 |
| Seattle | 26 |
| Berlin | 23 |
| Austin | 24 |
The London gap is real but smaller than it was. In the May 2026 snapshot, London (173) held a 34% edge over SF (129). By September, that margin is down to 26%.
Who is hiring in San Francisco?
| Company | SF roles | Signal |
|---|---|---|
| OpenAI | 75 | Accelerating |
| Perplexity | 16 | Accelerating |
| Plaid | 14 | Accelerating |
| Writer | 7 | Accelerating |
| Harvey | 4 | Steady |
| Deepgram | 3 | Accelerating |
| Anyscale | 3 | — |
| Hive | 3 | Steady |
The top four employers hold 112 of 163 roles — 69%. Compare that to New York, where Capital One (27) accounts for 25% of the market and no other single employer holds more than 13%. SF’s concentration is more acute.
Remove OpenAI and the remaining 88 non-OpenAI SF roles is a smaller pool than New York’s 106. The AI capital narrative runs on a single hiring pipeline.
What is OpenAI actually hiring for in SF?
OpenAI’s 75 San Francisco roles span five skill categories:
| Skill category | Roles |
|---|---|
| Applied AI | 20 |
| Data Scientist | 15 |
| AI Research | 11 |
| AI Software Engineer | 9 |
| Machine Learning Engineer | 7 |
Applied AI is the largest bucket — engineers building and deploying production systems, not publishing papers. The data mirrors OpenAI’s public direction: chip design, infrastructure scaling, enterprise and government deployment. Its tech signals include LLM, AI Infrastructure, Chip Design, pAGI, and Distributed Data Systems.
Top posted titles include Applied AI Engineer, AI Infrastructure Engineer, Research Engineer for Chip Design, and Applied AI Architect. The research roles are there, but they sit below applied deployment in volume.
What does the non-OpenAI SF market look like?
The 88 roles outside OpenAI are not a fallback pool — several of the employers are growing faster than OpenAI’s absolute numbers suggest.
Perplexity (16 roles) is in full build-out across ML infrastructure, search ranking, and agent capabilities. 80% of its total open roles globally are in SF. It added 4 roles between the July and September snapshots.
Plaid (14 roles) is the more surprising entry. A payments infrastructure company in the top three AI employers for San Francisco signals that the city’s AI market has a fintech layer most hiring conversations skip. Plaid’s AI buildout targets fraud detection, embedded insights, and ML infrastructure — senior-heavy roles, not junior pipelines.
Writer (7 roles) splits its hiring between SF and London, building enterprise generative AI and supporting the go-to-market expansion that comes with rapid commercial growth.
Harvey (4 roles) is a legal AI company: senior-weighted, SF-based, quiet on hiring volume but steady.
Which role types are in demand?
This is where SF diverges from the global index. Globally, Machine Learning Engineer leads all tracked role categories. In San Francisco, AI Software Engineer is first:
| Skill category | SF job count |
|---|---|
| AI Software Engineer | 59 |
| Machine Learning Engineer | 33 |
| Applied AI | 24 |
| Data Scientist | 22 |
| AI Research | 19 |
| AI Platform | 7 |
| AI Infrastructure | 6 |
| MLOps Engineer | 6 |
| AI Product Manager | 5 |
AI Software Engineer (59) outpaces Machine Learning Engineer (33) by nearly 2:1. Globally, that ratio is inverted. SF’s top employers — OpenAI, Perplexity — are past the research-heavy phase and hiring engineers who can ship working AI systems at scale, not primarily engineers who build new models from scratch.
Data Scientist roles (22) sit fourth. In London and Toronto, data science typically clusters around financial services. In SF it’s more diffuse — fraud at Plaid, product analytics at Perplexity, forecasting inside OpenAI’s operations.
How does SF compare on seniority?
| City | Total roles | Senior or above |
|---|---|---|
| San Francisco | 163 | 27% |
| London | 206 | 37% |
| Global index | 2,050 | 43% |
| Sydney | 71 | 51% |
| Berlin | 23 | 54% |
| Toronto | 121 | 61% |
San Francisco is the most mid-level market in the tracked index. 119 of 163 roles (73%) are categorized as mid-level — the highest share of any city. Only 27 roles are senior-heavy.
Two factors drive this. First, OpenAI’s SF operation skews toward applied engineering roles that many companies title at the mid-level IC band. Second, the SF startup layer (Perplexity, Harvey, Speak, Mach9) hires across seniority levels rather than filtering for senior talent only.
Candidates at the 3–6 year experience mark will find more surface area in SF than the city’s reputation for elite compensation and hypercompetitive culture suggests. 119 mid-level roles exist, most of them outside the frontier research track.
The 27 senior-heavy roles are concentrated at Plaid, Deepgram, and Harvey — companies operating in regulated or high-stakes verticals where production reliability outweighs headcount velocity.