# Research Engineer, Mid-Training

Hiring organization: [Cognition AI](https://career.thegoodapps.co/organizations/cognition-ai)

Canonical page: https://career.thegoodapps.co/jobs/b6c971a1-1885-407f-aef6-78cb5dedb9cd

Listed on Cognition AI's own careers site. Applications go to them directly.

- Seniority: Mid
- Location: San Francisco, CA

## Summary

This role focuses on optimizing the critical middle phase of LLM training, where raw capability is shaped into reasoning, generalization, and reliability. It suits researchers and engineers who excel at designing data strategies, training schedules, and synthetic pipelines—and who thrive making high-leverage decisions with minimal constraints.

_Our summary, not Cognition AI's wording._

## Skills named

Python, PyTorch, Statistics

## Required

- End-to-end familiarity with LLM training pipeline including pre-training, optimization, and architecture
- Hands-on experience with continual pre-training, annealing, or late-stage data mixing for large models
- Strong understanding of data quality and curation at scale
- Experience developing or evaluating synthetic data pipelines
- Proficiency in Python and deep learning frameworks like PyTorch
- Ability to debug distributed training systems at scale
- Strong fundamentals in optimization, statistics, and ML theory

## Nice to have

- Track record of original contributions through publications, open-source work, or internal results
- Experience extending context length in language models
- Familiarity with positional encoding strategies
- PhD or equivalent advanced training

Apply on Cognition AI's site: https://jobs.ashbyhq.com/cognition/57e1e2a1-e0a0-4448-a278-8eed035d7428/application