后端开发工程师（AI创新应用方向）-抖音
Core
Design and maintain self-developed large model training frameworks and optimize training/inference efficiency for Douyin's AI innovation tools.
Role type
Senior IC backend engineer (large model training infrastructure)
Builds
Self-developed large model training frameworks and AI innovation tools
Domain
AI/ML infrastructure, large language models, distributed systems
Deliverable
production ML models
Required skills
large model training frameworks, FlashAttention, PagedAttention, gradient checkpointing, mixed-precision training, high-concurrency service design, distributed systems, MySQL, NoSQL, message queues, data structures and algorithms
Preferred skills
experience leading key modules in large AI systems
Technologies
FlashAttention, PagedAttention, DeepSpeed, KVCache
Responsibilities
Maintain and optimize self-developed large model training frameworks; explore new interaction scenarios combining AIGC/generative models with live streaming, video, and social features; track industry frontier technologies to improve training and inference efficiency.
Seniority
Senior, hands-on IC