Who we'd reach at Featherless AIFound and reached by Power Apply
Best leadEngineering manager
Leads the team you would join
Tech lead
Leads the work day to day
Technical recruiter
Owns the opening
From your own Gmail and LinkedIn.
About the role
This is a full-time remote research role at Featherless AI, open worldwide, for a researcher who designs model distillation methods. The work turns large models into smaller, faster and more deployable systems while keeping quality, then takes the results into production with engineers. Publishing research is a core part of the role.
What you’ll do
- Design and evaluate distillation methods such as teacher-student training, self-distillation, layer-wise distillation and representation matching.
- Study trade-offs between model size, latency, memory and accuracy.
- Develop distillation approaches for large language models, long-context models and inference-constrained settings.
- Run large-scale experiments and ablations, and analyse results rigorously.
- Work with engineers to bring research outcomes into production.
- Write papers for top venues and contribute to notes, blogs and open-source projects.
What they’re looking for
- Strong background in machine learning research.
- Hands-on experience with distillation or related methods such as compression, pruning, quantisation or representation learning.
- Publication record, including conference, journal, workshop or arXiv work.
- Solid understanding of deep learning fundamentals.
- Fluency in PyTorch or an equivalent framework.
- Ability to communicate research ideas, results and limitations clearly.
Nice to have
- Experience distilling large language models.
- Work on efficiency research covering latency, memory or throughput.
- Experience with long-context or non-Transformer architectures.
- Open-source contributions in machine learning or research tooling.
- Prior startup or applied research experience.
What’s on offer
- Real ownership over research direction at a Series A company.
- Strong support for publishing and open research.
- Access to meaningful compute and production-scale problems.
- Remote working from anywhere, with a time zone range of UTC-11 to UTC+14.
Skills
Model distillationModel compressionPruningQuantisationRepresentation learningLarge language modelsDeep learningPyTorch
Questions about this role
Is this role remote?
Yes, the role is remote and open from anywhere in the world.
How much does it pay?
The salary is not disclosed in the posting.
Do I need publications?
Yes, the posting requires publication experience, which can include conference papers, journal papers, workshop papers or arXiv preprints.