Amy Yang Email & Phone Number

San Francisco, CA, US

Menlo Park, CA, US

GenAI (LLM, LDM) model post-training distributed inference performance optimizations with multi-GPU parallelization and low precision. Llama2&3, Emu model inference deployment and acceleration.

Menlo Park, CA, US

Performance optimizations for distributed GPU training and inference of ranking models.* Distributed multi-GPU inference for ads ranking models - service enablement, lower precision compute and embedding.* Accelerating multi-GPU training for ranking with quantized collective communications and low-precision embeddings.Publications:* Training Deep Learning.

Menlo Park, CA, US

Inference accelerator deployment for ads ranking models* Int8 post-training quantization and quantization-aware training for ads ranking models* Numerics research tooling and deployment monitoring for ranking inference acceleratorPublications:* Low-Precision Hardware Architectures Meet Recommendation Model Inference at Scale.

Scalable distributed key-value store as an alternative to blockchain for powering cryptocurrency and likewise resource management systems while maintaining decentralized trust, fault tolerance, anonymity, and protection against double spending.

Menlo Park, CA, US

Developed DB-based in-memory task queue for distributed static resources preprocessing of Facebook web services.

Boston, MA, US

Developed real-time chat service for native iOS mobile client; optimized responsiveness and memory footprint.