AI-Driven Engineering Diagnostics & Data Annotation Framework
Architected and managed an advanced AI-assisted Failure Analysis Management System (FAMS) to structure, label, and categorize massive engineering diagnostic datasets. • Data Annotation & Labeling: Structured and classified 9+ years of historical hardware failure logs, engineering summaries, and root-cause data into high-quality training datasets. • Model Evaluation & Optimization: Evaluated and aligned LLM semantic search models against 4M1E parameter logic, reducing data consolidation time from 4 hours to 15 minutes (94% reduction). • Workflow Prompt Engineering: Designed complex multi-shot prompting frameworks to eliminate model hallucinations in predictive technical risk alert pipelines.