{
  "basics": {
    "name": "YongBo Yu",
    "label": "AI Engineer and Data Scientist",
    "summary": "YongBo Yu, also known as Yong Yu, is a Toronto-based AI engineer and data scientist focused on agent systems, LLM workflows, and AI-native development with Codex. Built KiloDock, a multi-tenant gym operating system, and previously developed ML and ETL systems at Scotiabank supporting 1M+ users and $500K+ in estimated campaign savings.",
    "email": "yongbo9852@gmail.com",
    "location": {
      "city": "Toronto",
      "region": "ON",
      "countryCode": "CA"
    },
    "profiles": [
      {
        "network": "LinkedIn",
        "url": "https://www.linkedin.com/in/yongboyu"
      },
      {
        "network": "GitHub",
        "url": "https://github.com/YongBoYu1"
      }
    ]
  },
  "work": [
    {
      "name": "Scotiabank",
      "position": "Data Scientist",
      "startDate": "2022-08",
      "endDate": "2023-08",
      "summary": "Built an end-to-end insurance and mortgage propensity workflow for campaign audience prioritization on 1M+ user records, supporting an estimated $500K+ in campaign savings.",
      "highlights": [
        "Transformed 1M+ user records into model-ready features with SQL CTEs and window functions, cutting processing time 20%.",
        "Compared Logistic Regression, Random Forest, XGBoost, and LightGBM on imbalanced customer data using time-based validation and grid-search tuning, reaching a 0.857 validation F1 score and using SHAP to explain predictions.",
        "Delivered a serialized model and customer-level purchase probabilities for score-based campaign audience selection.",
        "Automated recurring model-development workflows on Airflow; built an occupation classifier that reduced manual scoring work by 50+ hours per month."
      ]
    },
    {
      "name": "Scotiabank",
      "position": "Data Scientist Intern",
      "startDate": "2022-01",
      "endDate": "2022-08",
      "summary": "Built decision-tree customer segments from behavioral data and automated campaign assignment.",
      "highlights": [
        "Reduced manual effort for monthly campaign launches by 51% through automation of customer segmentation and assignment.",
        "Delivered an automated data-generation and prediction pipeline for the finance team, reducing customer assignment time by 41%."
      ]
    }
  ],
  "projects": [
    {
      "name": "KiloDock",
      "startDate": "2025-10",
      "url": "https://yongbo-yu.vercel.app/projects/kilodock",
      "description": "CrossFit gym operating system (demo). Independently built a multi-tenant platform: React owner dashboard, React Native member app, FastAPI/PostgreSQL backend. Refined scheduling, booking, attendance, and workout-logging with two gym owners managing 200+ members.",
      "highlights": [
        "Automated booking, cancellation, check-in, and waitlist promotion in FastAPI using PostgreSQL row locks and server-side transactions to enforce membership and capacity rules.",
        "Reduced mobile schedule latency from 2.8–4.6s to 683ms and admin loading from six API calls/3.05s to one/778ms using aggregate SQL and gym-scoped caching.",
        "Built a Gemini-powered programming copilot using tool routing and 14-day workout history to detect repeated movement patterns and generate validated workout blocks for owner review."
      ]
    },
    {
      "name": "TradingAgents",
      "startDate": "2025-01",
      "endDate": "2025-10",
      "description": "Multi-agent trading framework. LangGraph workflow coordinating 10+ research, trading, and risk agents with conditional routing, stateful execution, and checkpoint recovery.",
      "highlights": [
        "Provider-agnostic LLM layer supporting OpenAI, Anthropic, Gemini, and Grok, with capability checks, reasoning controls, and dual-region routing.",
        "Pydantic structured outputs, verified market-data snapshots, instrument resolution, fallback data sources, and path-traversal protection."
      ]
    },
    {
      "name": "Enterprise RAG Agent Platform",
      "startDate": "2024-05",
      "endDate": "2024-11",
      "description": "Configurable RAG agent supporting three enterprise use cases through retrieval pipelines and business-specific logic.",
      "highlights": [
        "Reduced average response latency by 20% (to ~2s) with async streaming, embedding caching, and request queuing under simulated peak loads.",
        "Improved answer relevance by 25% and reduced hallucinations by 31% through chunking, reranking, and embedding optimization."
      ]
    }
  ],
  "education": [
    {
      "institution": "University of Victoria",
      "area": "Data Science",
      "studyType": "MS",
      "startDate": "2020-09",
      "endDate": "2022-12",
      "score": "3.7/4.0"
    },
    {
      "institution": "University of Prince Edward Island",
      "area": "Computer Science",
      "studyType": "BS",
      "startDate": "2016-08",
      "endDate": "2019-05",
      "score": "3.6/4.0"
    }
  ],
  "skills": [
    {
      "name": "Languages",
      "keywords": ["Python", "SQL", "TypeScript"]
    },
    {
      "name": "AI and LLM Engineering",
      "keywords": ["Multi-Agent Systems", "RAG", "LLM Evaluation", "Prompt Engineering"]
    },
    {
      "name": "Backend and Data",
      "keywords": ["FastAPI", "PostgreSQL", "Supabase", "Apache Airflow", "Pandas", "NumPy", "Feature Engineering"]
    },
    {
      "name": "Machine Learning, Frontend and Infrastructure",
      "keywords": ["XGBoost", "Decision Trees", "React", "Docker", "Chrome Extension APIs"]
    },
    {
      "name": "AI and agentic code tools",
      "keywords": ["Codex", "Claude Code", "Grok Build", "GLM 5.2", "LangGraph"]
    }
  ],
  "certificates": [
    {
      "name": "Tableau Desktop Specialist"
    }
  ]
}