Featherless AI is seeking a Machine Learning Engineer focused on AI architecture research to design and prototype next-generation neural network systems. This role operates at the intersection of research and engineering, transforming experimental ideas into scalable, production-ready models. The position emphasizes innovation beyond standard Transformer architectures, exploring new approaches to model efficiency, memory, and performance.
Responsibilities:
- Research and develop novel neural network architectures, including alternatives to Transformer models.
- Design and conduct experiments on scaling laws, memory mechanisms, and compute trade-offs.
- Build end-to-end model prototypes from research code to training-ready implementations.
- Collaborate with systems engineers to ensure efficient and deployable architectures.
- Analyze model behavior, failure modes, and inductive biases to improve performance.
- Review and extend cutting-edge research, contributing to internal knowledge and benchmarks.
Requirements:
- Strong foundation in machine learning and deep learning principles.
- Experience implementing model architectures from scratch.
- Proficiency with PyTorch or JAX for model development.
- Understanding of attention mechanisms, RNNs, state-space models, or hybrid systems.
- Knowledge of training dynamics, optimization, and scaling behavior.
- Ability to communicate complex architectural trade-offs effectively.
Benefits:
- Competitive compensation with equity participation.
- Opportunity to work on core AI architecture innovation.
- Collaborative environment with a high-caliber engineering team.
This role offers the opportunity to influence foundational AI model design in a research-driven startup.