{
  "schema_version": "1.0",
  "updated": "2026-09-09",
  "provenance": "Candidate-provided experience, edited for clarity. Not independently verified employment or performance data.",
  "identity": {"name":"Jay Kang","also_known_as":"Jianyu Kang","location":"New York, NY","email":"jianyu.mfe@gmail.com","linkedin":"https://www.linkedin.com/in/jaykang17"},
  "summary": "AI specialist with 10+ years across investment research, quantitative trading, and machine learning. Builds production data and AI systems, coordinates stakeholders, and owns delivery and post-launch support.",
  "interests": ["AI deployment and customer solutions","Product design informed by customer experience","Agent memory and retrieval","Spreading activation","Quantitative finance"],
  "preferences": {"work_priorities":["Own direction and strategy","Lead development","Build personally"],"preferred_location":"New York City","relocation":"Open for the right opportunity","travel":"Open to up to 50% travel"},
  "experience": [
    {"organization":"Mizuho","engagement":"External consultant via Synechron","role":"AI specialist consultant","start":"2025-12","end":null,"current":true,"scope":["Sole human developer using Copilot and coding agents","Stakeholder communication from business and data teams to the MD sponsor","Coordination of four human QA testers","Jenkins CI/CD, production incidents, and post-launch support"]},
    {"organization":"Alysida Capital","role":"Cofounder and head of quantitative trading and strategy","start":"2021-09","end":"2024-10","scope":["Hired and trained four researchers and developers; up to ten including summer interns","Hiring and dismissal authority as a partner","Model evaluation, trading simulation, leakage and overfitting review","Staged deployment, risk controls, investor and broker communication"],"scale_note":"Small firm managing proprietary and external capital; no performance claims provided here."},
    {"organization":"Synovation AI","role":"Director, machine learning and strategy systems","start":"2019-09","end":"2021-08","scope":["Financial machine-learning research frameworks","Reinforcement-learning environments for financial decisions"]},
    {"organization":"Meritz Asset Management","role":"Quantitative equity analyst","period_label":"Earlier career","location":"Seoul, South Korea","scope":["China-focused equity research","SVM-based universe selection","Company visits and fundamental research reports"]},
    {"organization":"Jichengli","role":"Intraday futures trader","start":"2009-10","end":"2012-01","scope":["Intraday commodity futures trading","Position monitoring and real-time market risk"]}
  ],
  "projects": [
    {"id":"governance","name":"Business and physical metadata platform","organization":"Mizuho","status":"Core platform in production; natural-language AI interface in beta","role":"Sole human developer and delivery owner using AI coding tools","adoption":{"active_domain_teams":4,"target_domain_teams":"10+","products":"10+","attributes":"Hundreds","enums":"Thousands","metadata_versions":"100+"},"capabilities":["Business-to-physical mapping through Unity Catalog","Draft-to-publish lifecycle with semantic versioning and approval gates","Domain entitlements and maker-checker controls","Audit records retaining changes and deleted inputs","Mapping coverage and integrity validation","Configurable data-quality metadata; 30+ standard rules plus custom rules"],"ai_design":"Sonnet and LangGraph handle natural-language queries and proposed changes. Consequential changes go through impact assessment and human confirmation before invoking application functions.","decision":"Launch the working Streamlit application first; defer an AppKit UI rebuild under schedule pressure.","stack":["Python","Streamlit","Databricks","Unity Catalog","PostgreSQL","LangGraph","Anthropic Sonnet","Jenkins"],"evidence_limits":["No measured time-saving percentage available","Data-quality rules are owned by the DQ team","AI interface is in beta"]},
    {"id":"parser","name":"Parser agent","organization":"Mizuho","status":"Production ingestion","role":"Built the complete agent and owns landing-to-bronze workflow","scope":{"datasets":10,"providers":1,"formats":"Approximately 5"},"workflow":["Read provider format documents","Generate Python parsers and loaders","Compare parsed document length with the original and retry mismatches through LangGraph","Human testing and review before deployment","Run approved loaders in daily Databricks Jobs"],"evidence_limits":["Document length checks do not establish semantic correctness","Demonstrated scope is one provider; no universal parser claim"]},
    {"id":"iris","name":"IRIS","status":"Personal experiment","focus":"Agent memory and retrieval using spreading activation","motivation":"Difficulty recovering useful context across AI interactions","evidence_limits":["Not a finished production memory product","No general performance advantage over RAG claimed","Source code is not included in the profile tools"]},
    {"id":"alysida","name":"Quantitative trading and research systems","organization":"Alysida Capital","status":"Prior professional work","scope":["Reusable model evaluation and simulation workflows","Leakage and overfitting review","Staged testing with small live capital","Golang execution and TimescaleDB data ingestion","Exposure controls and team development"],"evidence_limits":["No audited returns or institutional fund scale claimed"]}
  ],
  "education": [
    {"degree":"Master of Financial Engineering","institution":"Stevens Institute of Technology","year":2019},
    {"degree":"Master of Mathematics","institution":"Seoul National University","year":2014},
    {"degree":"Bachelor of Science in Mathematics","institution":"University of North Carolina at Charlotte","year":2009}
  ],
  "skills": {"core":["Python","SQL","PySpark","Databricks","Unity Catalog","PostgreSQL","Streamlit","LangGraph","Anthropic Sonnet","Jenkins"],"additional":["Databricks SDK/CLI/Jobs APIs","LangChain","JavaScript","HTML","Golang","AWS","TimescaleDB"],"languages":["English","Mandarin","Korean"]},
  "open_questions_for_interview": ["Commercial ownership of statements of work, budgets, procurement, and contract changes","Measured before-and-after deployment outcomes","Depth of security and operational evaluation beyond platform-provided controls","Exact dates for roles where a complete interval is not supplied"]
}