Job · Cognition AI
Research Engineer, Mid-Training
| Employer | Cognition AI |
|---|---|
| Location | San Francisco |
| Function | Data & research |
| Type | Full-time |
| Posted | 6 Aug 2026 |
About the role
Cognition AI is hiring a Research Engineer, Mid-Training to manage late-stage model training decisions. The role involves designing data mixtures, optimizing annealing schedules, and developing synthetic data pipelines to enhance coding and reasoning capabilities. Candidates need deep familiarity with the LLM training pipeline, hands-on experience with continual pre-training, and proficiency in Python and PyTorch.
Summary written by GAR, Octum’s cognitive engine, from the employer’s posting. The full description is on the employer’s site.
More roles at Cognition AI
Collected by Octum from Cognition AI’s careers site on 26 Oct 2026; Octum is not the employer and does not take applications. Spotted something wrong? Tell us.