AI训练数据与评测专家 - AI数据与安全
Core
Design training data strategies and evaluation frameworks for large language models across general capabilities, productivity, multimodal, coding, and agent scenarios.
Role type
Senior IC AI training data and evaluation specialist
Builds
Training datasets, evaluation benchmarks, and automated data production/assessment pipelines for seed base models and AI-native applications
Domain
Generative AI / Large Language Models / Data Engineering
Deliverable
production ML models
Required skills
Data mining, data construction, data cleaning and deduplication, annotation standard design, quality acceptance, distribution analysis, bias identification, long-tail scenario analysis, root cause analysis, automated pipeline design, prompt engineering, data synthesis, error classification, project management
Preferred skills
Background in finance, law, medicine, or education, experience with AIGC, multimodal models, agents, or code generation, strong AI tool usage habits
Technologies
LLMs, SFT, RL, RAG, Agent frameworks, Prompt engineering tools
Responsibilities
Design training data and evaluation schemes for specific business scenarios; Analyze negative cases and model errors to identify data or strategy gaps; Build automated workflows for data preprocessing, labeling, and quality reporting; Collaborate as a POC to clarify requirements and deliver projects; Standardize data methodologies, evaluation rubrics, and SOPs
