Senior ML Efficiency Researcher - Quantization & MoE

MLSyst 2020 is seeking a senior research role focused on making models efficient across training and inference. You will explore quantization, speculative decoding, distillation, mixture-of-experts, and sparse attention, balancing accuracy, latency, and cost. The work spans algorithm design, evaluation, and production-level deployment with the inference team. The position blends research depth with engineering impact, pairing model architecture decisions with practical production considerations