Research Engineer, Remote

Job Summary

To tackle the verification challenges of AI datacenters, the full-time Research Engineer will design challenge protocols, measure GPU performance under adversarial conditions, and collaborate with cryptographers and hardware-security experts in a remote environment.

Key responsibilities

• Design and implement challenge protocols to assess GPU performance and memory occupancy
• Conduct microbenchmarking and evaluate the feasibility of proposed protocols against real cluster topologies
• Collaborate with cross-functional teams to publish findings and address potential failures in protocols

Required qualifications

• Experience in GPU performance measurement or cryptographic protocols with a strong interest in hardware
• Rigorous approach to measurement and understanding of potential measurement pitfalls
• Ability to work in ambiguous environments and define problems collaboratively
• Familiarity with CUDA or equivalent low-level accelerator programming
• Knowledge of compression, information theory, or model quantization is a plus