Avatar of 黃文賢.
黃文賢
主任工程師 @ 群創光電股份有限公司 InnoLux Corporation
プロフィール
投稿
0つながり
印刷
Avatar of the user.

黃文賢

主任工程師 @ 群創光電股份有限公司 InnoLux Corporation
紹介文がまだありません。
Logo of the organization.
群創光電股份有限公司 InnoLux Corporation
Logo of the organization.
國立中央大學 National Central University
臺南市, 台灣
台湾

注目履歴書

アップロード日:12月 23日 2025

職歴・バックグラウンド

  • 現在の状況
    在職中
  • 専門分野
    Data Scientist
    Machine Learning Engineer
  • 業界分野
    人工知能/機械学習
    ビッグデータ
  • 職務年数
    15年以上 (15年以上 関連経験)
  • 管理経験
  • スキル
    Agentic AI
    Knowledge Graph
    Conversational UI
    Neuro-symbolic AI
    Causal Inference
  • 最終学歴

求職希望

  • 現在の状況
    新しい機会も検討中
  • 希望の雇用形態
  • 希望職種
  • 希望勤務地
  • フリーランス

職務経験

学歴

Logo of the organization.
博士号(PhD)
Computer Science and Information Engineering
2025 - 2029
説明
PhD Researcher / Graduate Research Assistant, National Central University, Taiwan Sep 2025 - Present Gen AI / Multi-Agent Systems / Knowledge Graph / Neuro-Symbolic AI/Causal Inference/NLP ● Enterprise Knowledge Graph (KG): Developed a high-precision KG for manufacturing control plans and design parameters, enabling cross-document retrieval and zero-hallucination querying for industrial standards. ● Predictive Diagnostics: Integrated Bayesian Networks with KG and LLMs to infer potential impacts on yield and reliability by reasoning through historical event correlations. ● Intelligent KM Copilot: Engineered a personalized Knowledge Management system with long-term memory and automated SWOT analysis, proactively guiding users with "Golden Rules" based on behavior patterns. ● Neuro-Symbolic Reasoning: Built a logic-based framework to verify design rule compliance; combined qualitative historical insights from KG to assess risks in real-time manufacturing events. ● Conversational AI & Auto-Scripting: Implemented a Text-to-Script engine that automatically generates and confi gures testing scripts based on natural language user requirements. ● Multi-Agent Hierarchy: Designed a Multi-Agent system using hierarchical planning to decompose complex queries into sub-tasks, featuring autonomous tool-use and self-correction for unexpected edge cases.