Careers at Pico
We're building open tools for systematic, hypothesis-driven language model research. If you're passionate about understanding how models learn, we'd love to hear from you.
Open Positions
ML Engineer
Remote · Full-time
Pico is a modular toolkit for systematic, hypothesis-driven language model research. We're looking for an ML Engineer to help build and scale the core infrastructure behind our training and analysis libraries — pico-train and pico-analyze.
What You'll Do
- Design and implement modular training pipelines for small language models (1M–1B parameters) with rich checkpointing and reproducibility features.
- Develop analysis tooling for studying training dynamics — convergence behaviour, gradient flow, effective rank, and representation geometry.
- Build distributed training infrastructure and optimize for efficiency across diverse hardware configurations.
- Contribute to open-source releases — models, datasets, and code — with a focus on reproducibility and community impact.
- Collaborate closely with researchers to translate hypotheses into rigorous, well-instrumented experiments.
What We're Looking For
- Strong experience with PyTorch and modern ML training frameworks.
- Familiarity with transformer architectures and language model pre-training.
- Comfort with distributed training (e.g. DeepSpeed, FSDP) and GPU cluster management.
- Interest in interpretability, training dynamics, or efficient training methods.
- Experience contributing to or maintaining open-source projects is a plus.
Interested? Send your CV and a brief note about what excites you about language model research to team@picolm.io.