多媒体算法工程师(AI Infra)-视频与边缘
Core
Develop and optimize algorithms for audio/video generation and multimodal large models, focusing on training efficiency, inference speed, and production deployment for business scenarios like video style transfer and smart voice interaction.
Role type
Senior IC machine-learning engineer (audio/video generation & multimodal models)
Builds
Production-ready audio/video generation and multimodal AI models
Domain
AI Infrastructure, Multimedia, Large Language Models
Deliverable
production ML models
Required skills
Python, PyTorch, FSDP, DeepSpeed, Megatron, CUDA, AscendC, Triton, TileLang, model quantization, parallel computing, TensorRT-LLM, SGLang, vLLM
Preferred skills
FlashAttention, Conv2d, Matmul, GroupedMatmul operator optimization, multi-modal algorithm principles
Responsibilities
Track and apply frontier technologies in audio/video generation and multimodal models; Optimize training frameworks and strategies to maximize hardware utilization; Optimize inference algorithms for speed and efficiency; Engineer model deployment from lab to production environments.