Back to Joseph Wang

Joseph Wang | Resume

AI Application Engineer / LLM Engineer / Agent Automation · Shanghai (Remote / Global)

joseph@josephjwang.com · +86 134 8223 9970 · linkedin.com/in/josephwang-ds

Download PDF ResumeSend EmailView LinkedIn

Professional Summary

9+ years in analytics, modeling, and automation, including 6+ years in machine learning and applied data science. Focused on turning LLM, RAG, and agent capabilities into deliverable business systems.

Core Capabilities

LLM Applications & System Design

  • Prompt engineering, function calling, tool use, and agent orchestration
  • Multi-model routing and failover across Qwen, DeepSeek, Claude, and GPT
  • RAG with hybrid retrieval (vector + keyword)
  • Evaluation design and hallucination-control optimization

Automation Platforms

  • Coze, Dify (ChatFlow / Workflow), n8n, MCP, Webhook/API integrations

Engineering & Data

  • Python (Pandas / Scikit-learn / Flask), SQL, Docker, Git, AWS basics
  • Visualization with Power BI, Tableau, and ECharts

ML & Analytics

  • XGBoost, LightGBM, Prophet, ARIMA
  • Pricing optimization, demand forecasting, segmentation, churn warning
  • NLP sentiment/topic pipelines and A/B testing

Professional Experience

Founder & AI Data Scientist

2022.09 – Present

Shanghai Yuyue Information Technology

  • Built an enterprise RAG FAQ assistant and improved answer accuracy to 90%+ while keeping hallucination under 5%.
  • Integrated Qwen, DeepSeek, Claude, and GPT APIs with reusable prompts, function calling, and tool chains.
  • Delivered multi-agent workflows in Coze and Dify for content and operations automation.
  • Shipped pricing optimization and demand forecasting models for inventory and growth decisions.

Pricing & Inventory Data Analyst

2020.07 – 2022.06

Lordco Auto Parts, Vancouver

  • Built pricing intelligence across 30,000+ SKUs and identified $2M+ profit opportunity.
  • Developed forecasting models for store-category planning and long-tail SKU replenishment.
  • Refactored ETL/reporting pipelines, reducing reporting time from 8 hours to 2 hours.

Business Analyst, Customer Insights & Aftersales

2017.11 – 2020.06

Mercedes-Benz Canada, Toronto

  • Built churn prediction and customer segmentation models with AUC 0.83.
  • Processed 50,000+ feedback records with NLP topic/sentiment analysis and improved NPS by 6 points.

Education

  • Northwestern University | M.S. Data Science
  • University of Waterloo | B.Math (Actuarial Science + Statistics)

Professional Training

  • LLM application training: API integration, function calling, RAG, and agent patterns (Workflow / ReAct / multi-agent).
  • Dify private deployment and AI workflow design.
  • AI-assisted engineering tools: Cursor, Claude Code, Codex, Trae, Tongyi Lingma.