Description
Tensordyne is hiring a Systems Performance Modeling Engineer to build and validate simulation and analytical models for multimodal generative AI inference workloads across compute, memory, collective communication, and network-fabric scales. The role involves developing trace-capture and replay tooling, modeling serving strategies and multi-hop fabrics, running calibration experiments on hardware, performing design-space analyses, and producing performance projections. Candidates should have hands-on experience with performance modeling, distributed ML execution, system architecture, and debugging prediction-measurement discrepancies, along with strong C++ and Python skills and an MS or higher in a relevant field.
