Machine Learning Researcher Springtail
Astera Institute is an active nonprofit research institute and foundation that operates in-house neuroscience-informed AI research and funds, operates, and supports open-science programs and public goods.
About Astera Institute
Astera operates three divisions: Neuro-AI, Radial life-sciences research, and its Foundation. Its AI work pursues neuroscience-informed approaches to artificial general intelligence, while its foundation and Radial activities support open scientific infrastructure, datasets, research programs, and residencies.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will research and develop architectures for data-efficient model induction, including bootstrapped program synthesis. You will hypothesize, test, and refine approaches to improve generalization and attention mechanisms, devise controlled datasets, investigate runtime inference and constrained optimization, and contribute to a documented, instrumented codebase.
Requirements
- Master's degree or equivalent in machine learning, mathematics, or an equivalent field
- Fluency with PyTorch
- Familiarity with JAX, CUDA, Triton, or their open-source ecosystems
- Demonstrated ability to conduct fundamental research
- Demonstrated ability to work in teams
Responsibilities
- Hypothesize, test, and refine approaches to improve generalization performance of architectural elements including attention
- Devise controlled datasets to study learning order and learned representations
- Investigate runtime inference in gradient-trained networks using statistical-learning and constrained-optimization approaches
- Contribute to a well-documented, well-instrumented, performant codebase