After OpenAI
- 20%Founded something8
- 17%Joined a startup7
- 17%Joined a big company7
- 46%Elsewhere
Shares of those who moved on, by their next stop.
People who leave OpenAI most often go to Anthropic, Thinking Machines Lab and Google. 20% found something next. 17% go small. Median stay 1y 5m
These are the engineers in the Zelcast network and the profiles run through the Eng Scorer, not OpenAI's headcount: 668 people on a map of 26,572, the No. 10 company on it by people. Updated August 20, 2026.
Put OpenAI side by side with another company, every number against every number.
Ribbon width is the number of people who made the move. Most arrive from Google. Most leave for Anthropic (26% of recorded departures).
Shares of those who moved on, by their next stop.
8 of the people who left went straight to a founding role, 20% of those who moved on, half of them within 2 months of leaving. The bars are the ’15 to ’26 classes; earlier founders count in the total.
668 stays with both a start and an end on the record.
2012 to 2026: 674 joined, 51 left.
there now 95 moved on 95. Current staff and alumni score about the same.
8 most common titles, by people who held them.
Everyone here with an Eng Scorer read, placed in their caliber tier.
664 of the 668 people here carry an Eng Scorer tier.
Exceptional and strong are the top two of the five tiers. 573 of the 664 people here with a read sit in them, 86%. A tier is the band an Eng Scorer caliber score falls in, so this is the same measurement as the median on the caliber card, opened up.
The level of the first title recorded here, against the last.
4%Titles that rose10 of 249 classified people
249 of the 668 people here had a title that could be placed on the ladder before and after.
A title rose when the last title recorded here sits higher on the ladder than the first; for someone who recorded one title here and then left, it is measured against the first title of their next role. The ladder runs intern, junior, mid, senior, staff or manager, principal, director, VP, then founder or C-level. A title that matches no rung is left out rather than guessed, and product, program and project managers are left out entirely, being a different ladder. That leaves 249 of the 668 people here classified, and it leans towards people who have left: someone still here on a single recorded title has nothing to be measured against.
The 8 most listed schools among the 662 people here with a school on file.
Of the 662 people here with a school on file, 115 list UC Berkeley, 105 list Stanford and 57 list MIT. A person can list more than one school, so these overlap rather than add up.
339 of the 664 people here whose profile was checked for public accounts. An account the check did not find reads the same as no account, so this is a floor rather than a census.
Across the 325 companies on the map that can show this figure, the median is 38%.
91% of the people who have worked at OpenAI and list a location live in the San Francisco Bay Area today, against 47% of everyone on the map.
664 of the 668 people who worked at OpenAI list a location, and every share here is over those 664.
Where they live now, from the profile as it reads today, not where they sat while they worked at OpenAI: a metro here does not mean the company has an office in it. The lower strip is the same metros across the whole map, for comparison.
Companies that show up most often alongside OpenAI, at any point in a career.
About this data: built from 144,786 jobs across 26,572 engineers in the Zelcast network and Eng Scorer reads, updated August 20, 2026. A move counts when the next role starts within 24 months of leaving; companies with fewer than 20 people on the map are not shown, and destinations with fewer than 3 people are folded into the totals. Caliber is the Eng Scorer score.
Worked at OpenAI? See how your profile reads to a startup hiring manager, and where it ranks among 26,572 engineers.
Arrow keys move along the map.