Zyphra
Research Engineer - Model Architectures
San Francisco · On-site · Full-time · Posted Mar 17, 2026
- Education
- Postgraduate degree in a scientific subject (Computer Science, EE/EECS, Math, Physics)
- Visa sponsorship
- Relocation and immigration support on a case-by-case basis.
About the job
Designs and rigorously tests novel model architectures and training methods to improve core modeling capabilities and address bottlenecks in contemporary models, collaborating closely with the pre-training team.
What you'll do
- Design and test novel model architectures and training methodologies.
- Investigate ways to improve modeling capabilities and address fundamental model bottlenecks.
- Work with the pre-training team to apply research insights to next-generation models.
What they ask for
- Strong research judgment and ability to take projects from conception through execution and write-up.
- Strong implementation and prototyping ability, including rapid experimentation.
- Ability to collaborate in a fast-paced research setting.
- High proficiency with PyTorch and Python.
Nice to have
- Experience with long-term memory, RAG/retrieval systems, dynamic/adaptive computation, or alternative approaches to credit assignment.
- Experience with reinforcement learning, control theory, signal processing, modern training pipelines, GPU-efficient architecture design, and rigorous experimental methods; prior ML publications are also noted.
Tools
PyTorch, Python
AI focus
Researches and develops novel AI model architectures and training methodologies, including work to improve modeling efficiency and capabilities.
Model-written summary · the original posting is authoritative