Your brand on every pageBecome a sponsor β†’

AI Researcher Jobs

AI research roles push model capability itself - pre-training, post-training, reinforcement learning, and the experiments behind frontier systems. This hub tracks titles explicitly labeled research scientist or research engineer; broader titles like 'member of technical staff' are excluded because they cover too many different functions to classify by title alone.

415 live AI Researcher roles across the AI employers we track - updated hourly, apply directly.

10new AI Researcher roles posted this week

AI Researcher roles are rare: 415 live right now. Get the new ones every Monday.

Get new AI Researcher jobs in your inbox

Join 100+ AI professionals Β· Weekly, free, unsubscribe anytime

Latest AI Researcher roles

Compute access defines the job

The research scientist and research engineer distinction has largely collapsed at frontier labs. It survives in job titles and sometimes in compensation bands, but the day-to-day converged because progress at scale is execution-bottlenecked rather than idea-bottlenecked: the person with the hypothesis needs to write efficient distributed training code to test it, and the person writing that code ends up making research decisions. Labs increasingly hire for both capacities in one person, and some drop the distinction entirely.

Compute access differs by orders of magnitude between employers advertising identical titles. A research scientist at a frontier lab may run experiments at a scale a university group or a small startup cannot approach, which changes not only the pace of work but which questions are askable at all. For candidates this is the most consequential and least advertised variable in an offer, and it is worth asking directly about cluster access, queue priority, and the size of a typical experiment.

Headcount follows the frontier, and the frontier has moved. Post-training work, covering reinforcement learning from human and automated feedback, reasoning and long-horizon training, tool use, and the data pipelines behind them, now absorbs more researchers than pre-training architecture work, which is concentrated in relatively few teams. Adjacent areas with growing demand include evaluation methodology, inference-time methods, and the systems research that makes large training runs finish.

Explore related searches

Frequently asked questions

How many live AI Researcher openings are there?
We are tracking 415 live AI Researcher roles across the AI companies we monitor, updated hourly. Each listing links straight to the employer's own application page.
Is a research scientist different from a research engineer at an AI lab?
Less than the titles suggest. At frontier labs both write substantial code, run experiments, and shape direction, because research at scale is limited by execution. Where the difference persists, scientist titles skew toward setting the agenda and publishing, and engineer titles toward infrastructure and experiment throughput. Read the responsibilities rather than the title.
How much does compute access vary between AI research employers?
By orders of magnitude. Frontier labs, well-funded startups, national compute programs, and university groups operate at very different scales, and the available scale determines which experiments are possible at all. Candidates should ask about cluster access, queue priority, and typical experiment size, because two identically titled roles can be very different jobs.
Which AI research areas are hiring most?
Post-training methods, including reinforcement learning from feedback, reasoning and long-horizon training, and tool use, together with the data work that supports them, absorb more headcount than pre-training architecture research. Evaluation methodology, inference-time methods, interpretability, and large-scale training systems research are also expanding.

Related AI job searches