Skill · AI Hiring Index · as of 2026-11-19
Fine-tuning in AI job postings
5% of technical postings at the AI companies we track mention Fine-tuning (256 postings). Research Engineer postings ask for it most: 23%.
Technical postings that mention it
5%
256 postings
Change since the previous Index
+1 pt
Which roles ask for it
Share of each technical role family's open postings that mention Fine-tuning.
| Role family | Share, as a bar | Share |
|---|---|---|
| Research Engineer | 23% | |
| Research Scientist | 19% | |
| AI / ML Engineer | 18% | |
| FDE / Applied | 8% | |
| Evals / Data | 3% | |
| Inference / Perf | 3% | |
| Safety / Policy | 3% | |
| Software Eng | 2% | |
| Infra / Hardware | 1% | |
| Security | 0% |
Companies asking for it most
| Company | Postings | Of its technical postings |
|---|---|---|
| Databricks | 29 | 5% |
| Mistral AI | 28 | 20% |
| Scale AI | 20 | 21% |
| Fireworks AI | 14 | 38% |
| Etched | 10 | 13% |
| CoreWeave | 10 | 6% |
| Anthropic | 8 | 2% |
| OpenAI | 8 | 2% |
| Liquid AI | 8 | 53% |
| Harvey | 8 | 9% |
How to learn it
- Hugging Face LLM Course free
Tokenizers, the Trainer, and fine-tuning chapters in runnable notebooks — enough to fine-tune a small open model on your own data.
- TRL documentation — Hugging Face free
The library behind most supervised fine-tuning and preference tuning (DPO, GRPO) you will be asked about, with working recipes.
Project: Fine-tune against a prompted baseline — LoRA-tune a small open model on one narrow task and compare it with a prompted frontier model on the same eval set — quality, latency and cost per 1,000 calls.