Summary

The Senior Applied Scientist at Nebius Token Factory researches and optimizes efficient LLM and VLM inference, developing methods and prototypes for production use. The role combines rigorous experiments and technical publication with close collaboration on deploying inference capabilities.

Responsibilities

• Lead research projects from hypothesis to production handoff
• Evaluate quantization and decoding optimization methods
• Build prototypes with PyTorch, Triton, and CUDA
• Design inference quality and performance evaluations
• Publish technical reports, papers, and open-source artifacts
• Collaborate with engineers to productionize research prototypes
• Mentor scientists on experiment design and system tradeoffs