In short
Across 1,766 live agentic AI listings, only 382 (21.6%) advertise remote work. The rate varies by specialisation, from AI infrastructure at 9.7% to generative AI at 31.6%. Across the four well-populated seniority tiers it barely moves, which contradicts the assumption that seniority buys location freedom.
The prevailing story about AI engineering is that talent is scarce enough to name its terms, and location is the first term you name. The listings do not support it.
Across 1,766 live agentic AI roles, 382 advertise remote work. That is 21.6%. Nearly four out of five require you to be somewhere specific.
Your specialisation matters more than your seniority
The aggregate hides a wide spread. Remote share by role category, for categories with at least 30 live listings:
- Generative AI: 25 of 79 roles, 31.6%
- AI Security: 13 of 43, 30.2%
- AI Solutions Engineer: 22 of 88, 25.0%
- AI Researcher: 17 of 72, 23.6%
- ML Engineer: 30 of 131, 22.9%
- General AI: 131 of 593, 22.1%
- AI Platform / SRE: 7 of 33, 21.2%
- LLM Engineer: 10 of 53, 18.9%
- Data Scientist: 15 of 86, 17.4%
- AI Product Manager: 5 of 32, 15.6%
- AI Agent Engineer: 22 of 142, 15.5%
- RAG / Search: 8 of 55, 14.5%
- Forward Deployed Engineer: 4 of 32, 12.5%
- AI Infrastructure: 19 of 196, 9.7%
The range runs from roughly three in ten down to one in ten. If remote work is a hard requirement for you, which specialisation you pursue is a bigger lever than any application tactic.
The AI infrastructure figure is the most reliable and the most consequential, because it rests on 196 listings, the second-largest category on the board. A great deal of the agentic hiring volume is concentrated in precisely the specialisation least likely to let you work from anywhere.
Seniority does not buy you location freedom
The assumption that you can negotiate remote once you are senior enough does not hold. Across the four tiers with enough listings to say anything reliable:
- Lead: 42 of 165 roles, 25.5% remote
- Mid: 190 of 800, 23.8%
- Senior: 80 of 396, 20.2%
- Staff: 53 of 332, 16.0%
That is a spread of about nine points across the entire middle of the career ladder, with the rate drifting down as you climb rather than up. Seniority buys compensation and scope in this market. It does not buy geography.
We are deliberately not reporting the junior, intern and executive tiers. Each currently holds fewer than 50 live listings, and when we measured them six days apart the junior figure moved from 22.9% to 6.7% and the executive figure from 9.3% to 28.9%. Those swings are sample noise, not signal, and any percentage quoted from them would be meaningless.
Where the remote roles actually are
The 382 remote listings spread across 202 companies, so this is not a story about three unusual employers. It is still lumpy at the top: the largest single remote employer accounts for 32 of them, with several others in the 14 to 16 range. Building a shortlist of consistently remote-friendly companies is a more efficient strategy than filtering a general board, because the same names recur.
How this was measured, and what it misses
These figures come from the location field of each live listing, matched for remote wording, across 1,766 roles on 9 August 2026. That method has a real limitation worth stating: it undercounts hybrid and quietly flexible roles. A job listed as a city name may still permit substantial remote work after negotiation, and would be counted here as non-remote.
So read 21.6% as the share of roles that advertise remote, which is the number that matters when you are filtering a board and deciding where to spend applications. The share that would ultimately accept a remote arrangement is certainly higher, and is not observable from listings.
One further caution, learned the hard way. Category percentages built on small listing counts move sharply week to week. Treat any figure here drawn from fewer than about 50 listings as directional, and weight the large categories, general AI, AI infrastructure, AI agent engineer and ML engineer, far more heavily than the small ones.
Frequently asked
What share of agentic AI jobs are remote?
21.6% of 1,766 live listings advertise remote work, which is 382 roles. That is based on the location field of each posting, so hybrid roles described by city name are counted as non-remote.
Which agentic AI specialisations offer the most remote work?
Generative AI leads at 31.6%, followed by AI security at 30.2% and AI solutions engineering at 25.0%. AI infrastructure is the clear outlier in the other direction at 9.7%, and it is one of the largest categories on the board.
Does being more senior improve your remote odds?
Barely. Across the four tiers with enough listings to be reliable, lead roles are 25.5% remote, mid 23.8%, senior 20.2% and staff 16.0%. If anything the rate declines as you climb.
Why are AI infrastructure roles so rarely remote?
The data shows the pattern but not the cause. Plausible explanations include proximity to hardware and datacentre operations, security requirements, and the concentration of these roles at a small number of large employers with strong onsite cultures.
Is remote work concentrated at a few companies?
Somewhat. The 382 remote roles spread across 202 companies, but the largest single remote employer accounts for 32 of them. Remote availability is broader than a handful of firms while still being unevenly distributed.