NYC AI Hiring Index · Issue #7 · November 14, 2026
Half of one week's postings were gone in a month. Two companies did it.
- Open roles
- 1,487across 47 companies
- Like-for-like change
- −2.9%47 companies, since Nov 4
- Roles in the city
- 782down 3, after a record 795
- Down within 28 days
- 23–49%nine weekly cohorts
This issue is late again: it was due on November 9 and covers ten days, November 4 to November 14. One thing in the data needed correcting along the way, and it is set out at the end.
1,487 open roles
Across 47 New York AI companies. 782 of those sit in the city itself and 302 are remote-eligible.
All 47 were tracked at both dates, so the raw and like-for-like figures are the same: open roles went from 1,532 to 1,487. Down 45, or 2.9%. The total held near 1,535 until November 9, then fell to 1,476 on November 13, the lowest since October 12.
13 companies grew, 20 shrank and 14 did not move. Two companies explain the fall. AlphaSense and ElevenLabs lost 40 roles between them, and the other 45 companies lost five. Among those, H1 and Aescape each lost seven roles and Andela six, while Ramp added 14, Attentive five and EvolutionIQ four.
The finding: one cohort came down twice as fast, and it was two boards
Every week we take the roles first seen that week and ask how many had come down four weeks later. For seven cohorts the answer sat between 23% and 37%. Two more cohorts closed their windows since the last issue:
| Roles first seen | Roles | Down within 14 days | Down within 28 days |
|---|---|---|---|
| Aug 16–22 | 115 | 13% | 25% |
| Aug 23–29 | 126 | 15% | 23% |
| Aug 30–Sep 5 | 108 | 13% | 31% |
| Sep 6–12 | 89 | 10% | 29% |
| Sep 13–19 | 118 | 12% | 27% |
| Sep 20–26 | 131 | 14% | 23% |
| Sep 27–Oct 3 | 121 | 23% | 37% |
| Oct 4–10 | 169 | 24% | 34% |
| Oct 11–17 | 184 | 27% | 49% |
90 of the 184 roles first posted in the week of October 11 were gone within four weeks. Issue six asked whether the eighth cohort would look like the seventh. It did, at 34%. The ninth went well past both.
It is not the market speeding up. ElevenLabs posted 54 of those roles and took down 45 of them inside 28 days. AlphaSense posted 38 and took down 21. Take those two boards out and the other 92 roles came down at 26%, which is where the first six cohorts sat. The same is true of the eighth cohort: 34% with them, 25% without.
So the range did not break. Two large boards are turning postings over much faster than everyone else, and they are big enough to move the average. From here the table needs reading with that in mind.
"Came down" means the posting disappeared: filled, cancelled or reposted under a new link, and we cannot tell those apart.
AlphaSense at its lowest, ElevenLabs near it
AlphaSense went from 202 open roles to 178, the fewest we have recorded. It had 241 on August 26. 52 postings came down in the ten days, a median of 30 days after going up, and 28 went up. Its New York roles finished where they started, at 35.
ElevenLabs went from 144 to 128. Its low is 122, on October 9. 58 postings came down, a median of 12 days after going up, and 42 went up, 18 of which were already gone by November 14. Its New York roles went from 12 to 10.
Issue six said ElevenLabs had given back more than a third of its October rebound. It has now given back nearly all of it.
The city: a record, then a second fall
Issue six asked whether the city would recover. It did, for three days. Roles in New York reached 795 on November 7, the most we have recorded, and then fell to 782.
| Oct 20 | Nov 4 | Nov 14 | |
|---|---|---|---|
| Roles in New York | 788 | 785 | 782 |
| Roles elsewhere | 764 | 747 | 705 |
Nearly all of this period's fall was outside the city: 42 of the 45 roles. But the city's small number hides a split. Ramp added 15 New York roles on its own. Without Ramp, the other 46 companies lost 18 roles in the city: Clay lost seven and H1 five, and Attentive added six.
EliseAI, which ended the city's run last time, is posting again: 16 new postings are still open. It also took down 20, open a median of 87 days, so its board went from 114 to 110 while its New York roles rose from 95 to 97.
Who is hiring hardest
| Company | Open roles | In New York | Change since Nov 4 |
|---|---|---|---|
| Ramp | 180 | 142 | +14 |
| AlphaSense | 178 | 35 | −24 |
| ElevenLabs | 128 | 10 | −16 |
| EliseAI | 110 | 97 | −4 |
| Rogo | 103 | 60 | +1 |
| Headway | 92 | 24 | −1 |
| Clay | 60 | 32 | −2 |
| Runway | 51 | 4 | +3 |
Ramp has the largest board in the index for the first time, at 180 open roles and 142 in New York, both its highest on record. In August it had 135. Rogo is at a record 103. Clay lost two roles overall and seven in the city.
Every open role at these companies is on the live jobs index, refreshed daily.
The middle of the ladder thinned
| Level | Now | Issue six |
|---|---|---|
| Mid | 758 | 792 |
| Senior | 297 | 319 |
| Staff and principal | 163 | 159 |
| Manager | 106 | 102 |
| Director | 61 | 64 |
| Entry | 59 | 53 |
| Leadership | 43 | 43 |
Mid-level and senior roles lost 56 between them, more than the whole index lost. Senior roles reached 292 on November 13, the fewest we have recorded. Entry level went the other way, from 53 to 59, after falling at the last issue. The routes in are in how to get into AI.
Go-to-market fell from 525 to 501 and engineering from 437 to 435, so the gap between them narrowed from 88 to 66. Engineering touched 426 on November 13, another record low. Design is at 40, also its lowest.
Forward deployed engineer roles went from 44 to 50, the most we have recorded, passing the 47 they reached on October 4.
What it pays
From salary ranges posted on New York listings. 91% of NYC postings carry one, up from 90%.
| Function | 25th | Median | 75th |
|---|---|---|---|
| Engineering | $187,500 | $215,000 | $252,500 |
| Product | $165,000 | $200,000 | $240,000 |
| Finance & Legal | $148,000 | $185,000 | $217,500 |
| Go-to-market | $140,000 | $175,000 | $217,000 |
| Operations | $130,000 | $155,000 | $180,000 |
Engineering's median is back at $215,000, on 154 posted ranges, after one issue at $216,000. Senior engineers stay at $207,500 and staff and principal at $250,000. The NYC AI engineer salary guide has the bands by level.
Go-to-market held at $175,000 on 157 ranges. Last issue we said its ten-thousand-dollar jump more likely reflected which sales roles were open than anyone paying more. It has now held for a second reading.
The tool counts: last issue's risers fell back
dbt went from 21 companies back to 18, where it was on October 20. TypeScript went from 38 to 35, Python from 40 to 38 and Go from 29 to 27. OpenAI went from 14 companies to 12, and Anthropic from 22 to 23.
These are small counts, and a few postings closing can move them. Every term, with who is asking for it, is in the AI skills glossary.
The top of the org chart
36 C-level, VP and head-of searches are open, up from 34. Five opened, five came down and two older postings joined the count.
Attentive opened two, a Chief of Staff, CEO and a VP, Strategy and Operations, and its Head of Applied Science / Data Science search came down after 86 days. Maven Clinic opened a VP, Consultant Relations search and ASAPP a Vice President, FP&A & Finance Engineering. ElevenLabs posted a Head of Music that came down the next day, the second head-of role in two issues it has posted and withdrawn within two days. Headway's Chief of Staff, Clinical search came down after 27 days.
A search coming down still does not tell you who, if anyone, was hired. Open searches are followed on the executive search tracker.
A correction to the data
On November 7, six of the companies we track had the web address of every one of their postings change, without the postings themselves changing. We identified a posting by its address, so each one was recorded as closed and reopened on the same day: 451 of them.
No published figure was affected, because it happened after issue six. Left alone it would have wrongly inflated this issue's time-to-fill table (the ninth cohort would have read 59%, not 49%), shown 14 executive searches opening and 13 coming down where five and five did, and reset the "days open" shown for those roles on the jobs pages.
We removed the 451 false closures, restored each role's real first-seen date, and now identify those postings by an ID that does not change when the address does. The November 7 row is marked as corrected.
Checking that turned up a second, older fault. Alloy's postings all share one web address, so we had been following its whole board as a single role. Its open-role counts were always right, but its postings never entered the time-to-fill figures and showed no "days open". They are now followed one by one, from the dates Alloy first published them, and stay out of the time-to-fill cohorts before today, because the Alloy postings that came down earlier were never recorded.
Both changes are on the methodology page.
Next issue: Monday, November 23. Whether AlphaSense and ElevenLabs keep shrinking, whether the city holds without Ramp, and the tenth cohort.
A note on method
Week-over-week changes are computed on the companies present in both weeks, never on the headline total. Adding a company to our coverage adds its whole board to the count and would show up as a hiring surge that never happened. We would rather report a small like-for-like change than a larger one that is really a bookkeeping change.
Read each morning from the job boards of the New York AI companies we track. Every company name links to their own application page — never ours. Full methodology and coverage on the index page.