AI数据平台开发工程师 - Seed Model
Core
Designing and evolving the architecture of a multimodal data platform to support the training and optimization of Seed's large-scale generative AI models.
Role type
Senior IC Data Platform Engineer (Multimodal/LLM)
Builds
Scalable multimodal data infrastructure for training large language models (MLLMs) and generative media.
Domain
Generative AI, Multimodal Data, Large Language Models
Deliverable
production ML models
Required skills
Python, Java, Data Modeling, Data Architecture, Structured/Unstructured Data Handling, Hive, ClickHouse, MySQL, MongoDB, Elasticsearch, Hadoop, Spark, Flink, Ray
Preferred skills
Experience building data platforms for large models or online/offline scenarios, experience with massive multimodal data platforms, open-source big data frameworks
Responsibilities
Propose data scale, type, and structure recommendations for model training; Build multimodal data platforms for storage, preprocessing, visualization, and labeling; Manage petabyte-scale multimodal data (video, images) including security and validation; Develop high-performance data pipelines capable of handling millions of QPS; Collaborate with algorithm teams to close the data loop for model evaluation and improvement.