// data.jsx — real content from Yuqi's CV + music identity.
// No fabrication. Everything here is from the actual resume/links.

const YUQI = {
  name: 'Jeff Ge',
  alias: 'JEdoubleF911',
  tagline: {
    en: 'Data scientist by day. R&B artist by night.',
    cn: '白天建模型，晚上写歌。',
  },
  location: 'Los Angeles, California',
  locationCn: '洛杉矶',
  email: 'jeff152152@gmail.com',
  phone: '+1 740 877 2267',
  linkedin: 'linkedin.com/in/jeff911',
  github: 'github.com/JeffGe911',
  music: 'untd.io/jedoublef911',

  summary: {
    en: [
      "I build data systems and ML pipelines for domains where decisions carry real consequences — currently focused on healthcare AI at xHealth Group, with prior work across urban analytics, real estate, and multi-agent LLM systems. USC MS Data Science (Spatial Track), graduating December 2026.",
      "My work starts before the model — scoping ambiguous problems, grounding decisions in business context, and designing systems that earn their place in production.",
      "Python · SQL · Spark · AWS · Airflow · XGBoost · LangChain · ArcGIS.",
    ],
    cn: '我把又脏又乱的空间数据，做成能真正落地的模型——房价、交通、风险、人。工具：Python / SQL / Spark / Tableau / AWS。工作之外，我写 R&B 和 hip-hop。',
  },

  // Experience — real, from resume, keyed for easy rendering
  exp: [
    { id:'idx', co:'IDXExchange', role:'Data Scientist Intern', period:'Jan — Apr 2026', loc:'Remote · Part-time',
      bullets: [
        'Led ML pipeline development to predict California single-family home close prices, finalizing XGBoost as the deployment model with R² = 0.8843 based on forward holdout validation stability and interpretability.',
        'Engineered geo-based and statistical features including SpatialLagPrice and Zip_MedianPrice using log-transformation and spatial lag techniques, improving model spatial accuracy.',
        'Lead implementing Linear Regression, Random Forest, LightGBM, XGBoost, and Stacking using R², log R², MAPE, and MdAPE.',
        'Benchmarked 5 models (Linear Regression, Random Forest, LightGBM, XGBoost, Stacking) across R², MAPE, and MdAPE; led model selection discussion with 3-person team.',
        'Proposed future roadmap including luxury vs. non-luxury market segmentation and price calibration strategies.',
      ] },
    { id:'unim', co:'UNIM Innovation', role:'Data Engineering Intern', period:'Jun — Aug 2025', loc:'Shanghai · On-site · Full-time',
      bullets: [
        'Designed and deployed end-to-end ETL pipeline (AWS + Airflow + DataHub) consolidating manufacturing demand data across 50K+ samples from disconnected sources.',
        'Built time-series forecasting models (Prophet + XGBoost ensemble) improving demand forecast accuracy by 35% over baseline.',
        'Automated reporting workflows reducing manual reporting time by 80%, enabling weekly stakeholder updates instead of monthly.',
      ] },
    { id:'aur', co:'Aurite AI', role:'AI Agent Builder', period:'Jun — Aug 2025', loc:'Remote · Part-time',
      bullets: [
        'Designed and deployed multi-agent AI system for institutional-grade portfolio management.',
        'Built User Preference & Portfolio Optimization agents (LangChain + LLM APIs).',
        'Collaborated with cross-functional team to integrate market analysis and personalized asset allocation.',
      ] },
    { id:'ica', co:'Icarus Fund LLC.', role:'Equity Research Intern', period:'Jun — Sep 2024', loc:'NYC · On-site · Full-time',
      bullets: [
        'Built an ML-driven equity watchlist in Python — combining factor screening, momentum signals, and fundamental filters to surface high-conviction candidates from the broad market.',
        'Performed DCF valuations and comparable company analyses on watchlist names; presented investment theses to the team to inform portfolio positioning.',
        'Optimized SQL queries on the market data warehouse, reducing daily pipeline runtime by 40% and unblocking faster research workflow.',
      ] },
    { id:'evp', co:'EVERPRO Insurance Brokers', role:'Data Analyst / Junior Broker', period:'Jun — Aug 2023', loc:'Shanghai · On-site · Full-time',
      bullets: [
        'Researched optimal cash values across firms.',
        'Designed whole life insurance plans in Excel + PowerPoint.',
        'Signed two contracts totaling 1,000,000 RMB in premiums.',
      ] },
    { id:'tow', co:'Towers Consulting', role:'Business Consultant Intern', period:'May — Aug 2022', loc:'Remote · Full-time',
      bullets: [
        'Conducted firm performance analysis and competitive benchmarking to support client recommendations.',
        'Led supply-chain stakeholder outreach, achieving 50% response rate.',
        'Built interactive Tableau dashboards visualizing real-time product metrics for client teams.',
      ] },
    { id:'kea', co:'Kearney', role:'Business Analyst Intern', period:'May — Aug 2021', loc:'Shanghai · On-site · Full-time',
      bullets: [
        "Supported Project Shark, a cross-border M&A advisory engagement on fishmeal procurement, in collaboration with Kearney's Germany office.",
        'Conducted primary research and built market intelligence materials on industry suppliers through targeted stakeholder outreach.',
      ] },
  ],

  edu: [
    { school:'University of Southern California', deg:'M.S. Data Science (Spatial Track)', dates:'Expected Dec 2026', cur:true,
      coursework:['Machine Learning', 'Deep Learning', 'Big Data Computing', 'Spatial Statistics', 'GIS Programming', 'Database Systems'] },
    { school:'University of Wisconsin–Madison', deg:<>B.S. Economics<br/>B.S. Personal Finance</>, dates:'Graduated Aug 2024',
      coursework:['Econometrics', 'Statistical Methods', 'Financial Modeling', 'Health Economics'] },
  ],

  skills: {
    Programming: ['Python', 'SQL'],
    'ML / AI': ['XGBoost', 'LightGBM', 'scikit-learn', 'LangChain', 'OpenAI API', 'Anthropic API'],
    'AI Workflow': ['Claude Code', 'Cursor', 'GitHub Copilot'],
    'Data Engineering': ['Spark', 'Airflow', 'pandas', 'SQLite'],
    Spatial: ['ArcGIS Pro', 'GeoPandas', 'H3'],
    Cloud: ['AWS'],
    Statistics: ['Hypothesis Testing', 'A/B Testing', 'Regression', 'Time-Series', 'Spatial Statistics'],
    'Viz / BI': ['Tableau', 'Power BI', 'matplotlib', 'ArcGIS Story Maps'],
    Spoken: ['Mandarin · Native', 'English · Native'],
  },

  certs: [
    'Data Science Foundations: Python Scientific Stack (CoderPad)',
    'Data Cleaning in Python Essential Training',
    'Business Etiquette: Phone, Email, and Text',
  ],

  now: [
    { label: 'Finished',  text: 'LA traffic crash hotspot prediction (H3 + XGBoost) — SSCI 575', link: 'https://arcg.is/CaHe93' },
    { label: 'Upcoming',  text: 'Summer Engineer @ xHealth Group (Summer 2026)' },
    { label: 'Studying',  text: 'H2O: Heavy-Hitter Oracle for Efficient Generative Inference of LLMs (NeurIPS 2023)' },
    { label: 'Next',      text: 'Final semester at USC MSDS · Graduating Dec 2026' },
  ],

  // Music tracks — links go to the UM profile page; real track names to come.
  tracks: [
    { n: '01', title: 'Our Songs',  len: '', tag: 'R&B',     year: '2025', slug: 'our-songs' },
    { n: '02', title: 'Passing On', len: '', tag: 'R&B',     year: '2024', slug: 'passing-on-1' },
    { n: '03', title: 'Thoughts',   len: '', tag: 'hip-hop', year: '2023', slug: '6463b137dd60f36c1fb9b52f' },
    { n: '04', title: 'Double',     len: '', tag: 'hip-hop', year: '2022', slug: '633d88a650866779e53b8aed' },
  ],
};

Object.assign(window, { YUQI });