Which roles ask for RAG at tech companies · 687 roles · 90 days to Nov 21, 2026
Updated
687 roles counted that list RAG, one per company, title and level; 1,120 postings in the window.
In short
RAG: listed in 687 roles at tech companies whose requirements we read, in the 90 days to Nov 21, 2026. It was required in 438 and preferred in 249. The largest group is AI engineer roles: 208 of 687 (30%).
- Required
- 438 of 687 (64%)
- Of the roles that list RAG
- Preferred
- 249 of 687 (36%)
- Of the roles that list RAG
- Largest group
- 208 of 687 (30%)
- AI engineer
Roles that ask for RAG
Roles listing RAG, by title family. Where the role has a page, the count is also out of that role's roles whose requirements we read.
| Role | Roles | Of the role's roles we read |
|---|---|---|
| AI engineer | 208 | 208 of 507 (41%) |
| Software engineer | 169 | 169 of 4,983 (3%) |
| Machine learning engineer | 73 | 73 of 538 (14%) |
| Forward deployed engineer | 68 | 68 of 340 (20%) |
| Data scientist | 58 | 58 of 773 (8%) |
| Data engineer | 44 | 44 of 867 (5%) |
| Security engineer | 18 | 18 of 801 (2%) |
| Data architect | 12 | 12 of 126 (10%) |
| Platform engineer | 10 | 10 of 454 (2%) |
| Software architect | 10 | 10 of 207 (5%) |
Companies that ask for it most
- Guidehouse6 roles
- Tiger Analytics6 roles
- Bask Health5 roles
- AJAIA4 roles
- CACI International Inc4 roles
- Databricks4 roles
- Lifelancer4 roles
- Robots & Pencils4 roles
- Robots and Pencils4 roles
- Snowflake4 roles
Asked alongside
Out of the 687 RAG roles whose requirements we read: the other skills they ask for. At most 10 per list, those asked for most.
| Skill | Roles asking |
|---|---|
| Python | 485 of 687 (71%) |
| LLMs | 420 of 687 (61%) |
| AI | 346 of 687 (50%) |
| AI Agents | 254 of 687 (37%) |
| AWS | 247 of 687 (36%) |
| Machine Learning | 212 of 687 (31%) |
| Prompt Engineering | 208 of 687 (30%) |
| CI/CD | 204 of 687 (30%) |
| SQL | 175 of 687 (25%) |
| Azure | 157 of 687 (23%) |
| Skill | Roles asking |
|---|---|
| AI Agents | 129 of 687 (19%) |
| Vector Databases | 120 of 687 (17%) |
| LLMs | 107 of 687 (16%) |
| AWS | 96 of 687 (14%) |
| Docker | 86 of 687 (13%) |
| CI/CD | 76 of 687 (11%) |
| LangChain | 72 of 687 (10%) |
| Kubernetes | 68 of 687 (10%) |
| Observability | 67 of 687 (10%) |
| AI | 66 of 687 (10%) |
Experience
Asked most often: 5–7 years, 242 of 533 (45%).
Of the 533 roles that state years
- Under 1 year
- 6 of 533 (1%)
- 1 year
- 15 of 533 (3%)
- 2 years
- 32 of 533 (6%)
- 3–4 years
- 99 of 533 (19%)
- 5–7 years
- 242 of 533 (45%)
- 8–9 years
- 69 of 533 (13%)
- 10+ years
- 70 of 533 (13%)
Degree
122 of 680 (18%) require a degree, and 132 more accept experience instead.
Of the 680 roles we could read for it
- Required
- 122 of 680 (18%)
- Required, or experience instead
- 132 of 680 (19%)
- Preferred
- 47 of 680 (7%)
- Not mentioned
- 379 of 680 (56%)
Pay and remote work
Of the 687 roles we could read for it. Pay these roles state, not what the skill pays.
- Median stated pay
- $170k a year
- Roles stating pay
- 295 of 687 (43%)
- Remote roles
- 576 of 687 (84%)
How we count: 687 roles from 549 employers in the 90 days to Nov 21, 2026, each role counted once however many boards post it. Skills are counted out of the roles whose requirements we read. Roles are not filtered by country. Methodology
Something wrong here?