Mechanistic analysis of how neural networks compute — circuit tracing, attribution, behavior decomposition. The research field underpinning safety arguments about model internals.
Primary AI lifecycle stage: evaluation.
As of today, 82 active AI roles across 31 companies in our index reference Interpretability. Hiring concentrates at the post-training (51%) and agents (17%) stages. Most common sectors: Big Tech, AI Frontier, Banking. New postings fell 27% in the last 30 days versus the prior 30 (30 → 22).
9 AI roles tagged interpretability.
| Company | Title | Sector | AI score | Other tags |
|---|---|---|---|---|
| Capital One | Applied Researcher II, AI Foundations | Banking | 9 | Fine-tuning · Frontier research · RLHF · Vector DB |
| Capital One | Applied Researcher I | Banking | 8 | Pretraining · Fine-tuning · Frontier research · RL post-training · Vector DB · Model serving |
| Capital One | Applied Researcher II | Banking | 8 | Fine-tuning · RL post-training · Vector DB |
| Capital One | Applied Researcher I | Banking | 8 | Fine-tuning · Frontier research · RLHF · Vector DB · Recommender systems · Model serving |
| Capital One | Applied Researcher II | Banking | 8 | Fine-tuning · RL post-training · Recommender systems · Vector DB |
| Capital One | Applied Researcher II | Banking | 8 | Fine-tuning · RL post-training · Vector DB · Model serving · Recommender systems |
| Capital One | Applied Researcher I (AI Foundations) | Banking | 8 | Pretraining · Fine-tuning · Vector DB · Frontier research · RLHF |
| Capital One | Principal Associate, Data Scientist - Retail Bank | Banking | 7 | |
| JPMorgan Chase | Wealth Management, Quantitative Portfolio Manager, Equities CIO | Banking | 5 |
Mechanistic analysis of how neural networks compute — circuit tracing, attribution, behavior decomposition. The research field underpinning safety arguments about model internals. Primary AI lifecycle stage: evaluation.
82 active AI roles across 31 companies in our index reference Interpretability as of today. New postings fell 27% in the last 30 days versus the prior 30 (30 → 22).
The companies with the most active Interpretability listings are: Microsoft (9 roles), OpenAI (8 roles), Amazon (7 roles), Meta (7 roles), Anthropic (6 roles).
Interpretability primarily belongs to the evaluation stage of the AI lifecycle. In current hiring, Interpretability roles concentrate at: post-training (51%), agents (17%).
The sectors with the most active Interpretability hiring are: Big Tech, AI Frontier, Banking.