Lead research efforts exploring new neural network foundations and architectures, going beyond incremental improvements
Advance neural networks broadly, with a particular focus on LLMs as a key application area
Rethink model representations and computational primitives
Explore hypercomplex neural networks and alternative mathematical formulations
Investigate analog, mixed-signal, and custom hardware approaches
Design and run experiments and prototypes to validate novel hypotheses
Define evaluation methodologies and compare new approaches against state-of-the-art baselines
Translate research ideas into scalable implementations and measurable results
Collaborate closely with engineering to bridge research and practical systems
Act as the technical lead for this research direction
Collaborate with the ML engineering team to support complex ML engineering projects by providing cutting-edge insights and guiding key technical decisions
What we look for
Strong background in machine learning research or advanced ML engineering
Deep understanding of modern deep learning architectures and their limitations
Proven ability to formulate original ideas, design rigorous experiments, and iterate based on results
Strong curiosity about fundamental ML questions, not just applied ML
Professional experience training or fine-tuning frontier models; extensive hands-on personal projects are also acceptable
Hands-on experience with reinforcement learning (RL), including areas such as RLHF and policy optimization
Comfortable working in ambiguous, open-ended research environments while maintaining a strong focus on outcomes, prioritization, and rapid validation of ideas
Strong plus
Experience optimizing large-scale inference systems (latency, throughput, memory efficiency), including practical understanding of memory movement, KV cache behavior, and quantization
Experience with hardware-aware ML or hardware design