Kebei Jiang

About Kebei Jiang

Helped/led the developing of end-to-end knowledge solutions in energy domain based on Azure Cognitive Search; experienced with building indexing pipeline, Lucene query syntax and on-boarding custom enrichment skills.• Developed multiple deep-learning/rule-based functionalities, ranging from image classification to NER; experienced with the full cycle of ML development, ranging from data collection to deploying models as web services.• Experienced with production tools such as FastAPI, Unit Testing, Azure DevOps and ARM Template.• Experienced with DS frameworks such as Pytorch, spaCy and Huggingface

Experience

1. Senior Applied Scientist

   Microsoft

   Jan 2020 — Present · Seattle, WA, US

   Experienced with LLMs (e.g, GPT and Llama) and VLMs (e.g, Qwen2-VL)- prompt engineering and synthetic data generation using LLMs - alignment with skills such as SFT, DPO and GRPO - domain/skill adaptation with SFT- Experienced in document knowledge mining - document ingestion, indexing/search, skill enrichment and evaluation - building end-to-end RAG pipeline and performance enhancement- Experienced with the full cycle of ML development, ranging from data processing to model deployment

Education

• South China University of Technology

  Bachelor's degree, Physics

  2003 — 2007

• Louisiana State University

  Doctor of Philosophy (Ph.D.), Physics

  2007 — 2014

Skills

• Statistics
• Science
• Physics
• Matlab
• Research
• C++
• Java
• Linux
• Student Affairs
• Quantum Optics
• Python
• Latex
• Teaching
• Higher Education
• Mathematica