attention head AI jobs Alerts Saved
AI jobs › Pre-training AI jobs

Pre-training AI jobs

Pre-training is where a foundation model learns most of what it knows: a large network trained on vast amounts of text, code, images or other data, usually to predict what comes next. A run takes a large cluster of accelerators for weeks or months, so mistakes are expensive, and much of the work is making sure a run is worth doing before it starts.

Research scientists choose the architecture, the data mix and the training recipe, and study scaling laws: how a model improves as it gets bigger, sees more data or uses more compute. Research engineers and training-infrastructure engineers keep large runs fast and stable, with parallelism across many devices, checkpointing, and watching for loss spikes and hardware failures. The work sits mostly at AI labs and companies that train their own models, and expects a deep understanding of transformers, PyTorch or JAX, and distributed training.

We count roles whose postings mention pre-training or pretraining, scaling laws, or foundation model training. Pretrained counts too, and many applied roles say it only because they build on pretrained models, so not every role here trains one.

154 open AI roles at 74 employers, from their own job boards. Showing the newest 100.

Follow Pre-training6 new this week

New Pre-training roles by email, the morning after we find them, in one email with your other alerts. Narrow it, or keep every role.

How often
We email a link to confirm it.
NVIDIA · 10Reflection AI · 10Anthropic · 7Hark · 6Waymo · 6Amazon · 5Mistral AI · 5Cursor (Anysphere) · 4Pluralis Research · 4Rhoda AI · 4Thinking Machines Lab · 4Aleph Alpha · 3Applied Intuition · 3Articul8 · 3Cerebras · 3

Get new Pre-training roles every Monday

A job alert: new Pre-training roles from employers' own job boards, every Monday. Unsubscribe in one click.

Related