Senior Data Scientist
Nearly two decades building and validating statistical models at CDC and NIOSH. Now building detection, credit-scoring and agentic AI systems at OhMyCode LLC on AWS, Snowflake and Databricks.
A Georgia-based data and AI company founded in 2026. We build systems that make public data verifiable โ every figure traceable to its original source.
Electoral Insights Hub โ bilingual analytics for Korea's 18thโ21st presidential elections and U.S. elections, built on open tally-sheet archives. Currently hosting the 18thโ21st Election Data Competition.
MacroPulse AI โ quant-finance tools on Amazon Alexa+ (regime detection, Monte Carlo VaR, FOMC stress tests) โ and LLM agents whose numbers are verified, not asserted.
Non-partisan and transparent by design. Archives, methods and verification steps are published so anyone can reproduce a result from the originals.
Founded 2026 ยท Duluth, Georgia, USA ยท Founder & CEO HeeKyoung Chun
60์ด ์๊ฐ ์์ โ 4๋ง ๋ฒ์ ์๋ฎฌ๋ ์ด์ ์ผ๋ก 2026 ์ค๊ฐ์ ๊ฑฐ๋ฅผ ์์ธกํฉ๋๋ค
Production applications built and operated by OhMyCode LLC โ click any card to try it live.
Analyze Korea's 18thโ21st presidential election data using the original tally-sheet archives. Verify with the originals, not with claims. First prize $1,000 โ entries close Dec 31, 2026.
Bilingual (EN/KR) U.S. election analytics โ turnout, vote share, swing, margin and recount views, plus a Monte Carlo Senate forecast with uncertainty bands.
Bilingual (KR/EN) analytics for Korea's 18thโ21st presidential elections โ vote share, swing and turnout across all 17 regions, 2012โ2025.
Investment dashboard visualizing markets and economic indicators with risk flags โ Monte Carlo VaR, regime detection and FOMC stress tests.
Educational demo: LLM-generated reports with automated number verification and hands-on exercises.
Senior Data Scientist with a Doctor of Science degree and nearly two decades transforming complex healthcare data into actionable insights at CDC and NIOSH, now applied to detection, scoring and agentic AI at OhMyCode LLC.
Bridging technical analysis and executive decision-makingโtranslating statistical results into recommendations leadership can act on.
Detection and scoring systems, healthcare analytics, RAG systems and AI agents
RAG pipeline: PDF ingestion into ChromaDB, hybrid BM25 + dense search with Reciprocal Rank Fusion, FastAPI backend, RAGAS evaluation on a synthetic golden dataset. Deployed via Docker / Cloud Run.
Autonomous LLM agent that plans multi-step research tasks, queries web sources, synthesizes findings, and generates structured reports with citations.
Automated pipeline to transcribe, summarize, and surface insights from HK podcast episodes using speech-to-text and LLM summarization, enabling searchable audio knowledge bases.
End-to-end scoring pipeline with reject inference implemented three ways and scored against ground truth. Found and fixed four deployed defects, including a calibration bug that moved scores 45.6 points.
Python port of a CDC production R/SAS pipeline: six-source weekly merge, change-point alert kept distinct from a magnitude threshold, completeness and cross-source reconciliation checks.
Local AI coding agent with Splunk cost, latency and DLP observability; detected anomalies trigger self-healing model-router re-weighting. 6th place, Splunk Dashboard Contest 2026.
AWS-hosted MCP server exposing regime detection, Monte Carlo VaR and FOMC stress tests to Amazon Alexa+, backed by a 14-tab Streamlit dashboard.
Rule engine for ICD-10โHCC mapping and NDC drug-interaction checks, paired with a tool-calling agent that acts on rule outputs.
Multi-provider LLM layer on AWS Bedrock (Nova Pro, Converse API) and Gemini, with head-to-head scoring of latency and output quality on identical prompts.
Predictive models and NLP for CDC surveillance data
Entity extraction using Spark NLP
Propensity score models for policy analysis
CI/CD for reproducible ML workflows
2025โPresent
OhMyCode LLC
Detection and alerting pipelines, credit scoring with calibration QA, and agentic AI on AWS, Snowflake and Databricks.
2022โ2025
CDC (DNI Contractor)
Lead predictive and causal models. Advanced NLP analyses.
2019โ2022
CDC (Goldbelt, TJFACT)
NVDRS and COVID-19 surveillance analytics.
2008โ2017
CDC, NIOSH
Cost-effectiveness modeling for prevention programs.
2014โ2020
Georgia Southern University
Biostatistics and epidemiological modeling research.
"A cross-sectional study of teens' health skills that were enhanced through school health education and their associations with mental health-related outcomes" โ Health Education & Behavior
Accepted ยท forthcoming 2026
"Adolescent and Parent Perspectives on Sexual and Reproductive Health Information: A Cross-Sectional Study of Perceived Usefulness and Comfort" โ Journal of Adolescent Health
Accepted ยท forthcoming 2026
"Evaluation of the Reliability and Validity of the Perceptions of Skills Enhanced Through School Health Education (PSE-SHE) Measure" โ Journal of School Health
doi.org/10.1111/josh.70038
"Updated assessment of occupational safety and health hazards of climate change" โ Journal of Occupational and Environmental Hygiene, 20(5โ6):183โ206
doi.org/10.1080/15459624.2023.2205468
"Maternal exposure to air pollution and risk of Autism in children: A systematic review and meta-analyses" โ Environmental Pollution
doi.org/10.1016/j.envpol.2019.113307
JNC TV is a non-profit Korean-American civic media organization based in CA. In addition to community journalism, JNC TV operates a non-commercial civic data research program that turns official public records โ national health-survey data and election records โ into verified, reproducible, bilingual (KoreanโEnglish) public resources for the Korean-American community and the wider public.
All research conducted under this program is non-commercial. Data, code, and findings are published openly, and every published figure is independently verified against its source records before release.
Principal Investigator / Scientific Lead
Dr. Chun leads the design, analysis, and verification of all research conducted under JNC TV's civic data research program. She was a survey statistician and senior data scientist in health analytics at the U.S. Centers for Disease Control and Prevention (CDC), specializing in complex survey design and weighting, trend analysis, and reproducible statistical pipelines. Her current work includes a re-analysis of U.S. Youth Risk Behavior Survey (YRBS) microdata, 2015โ2025, and the statistical validation of official election tally records.
Contact: tcgyver@jnctv.com
Research Engineer / Data Product Lead
Sechan Tak is a member of the JNC TV staff and builds the data products and content pipelines for the research program, including interactive dashboards and weekly public-data monitoring reports. He is a graduate of Vanderbilt University (Political Science and Economics, Computer Science minor).
Research Collaborator
Dr. Song, a research scientist at Vanderbilt University, collaborates with the program on election-data research, including the bilingual Electoral Insights Hub, an open analytics platform for Korean and U.S. election data.
A fully reproducible re-analysis of public CDC Youth Risk Behavior Survey (YRBS) microdata examining whether the post-2021 improvement in adolescent mental health and risk behaviors reached all sex and racial/ethnic groups. Outputs: peer-reviewed manuscripts and public bilingual summaries.
Digitization and statistical validation of 18,949 official Korean ballot tally sheets, linked to precinct-level analysis, together with a calibrated Monte Carlo forecasting model for the 2026 U.S. midterm elections (40,000 correlated-error simulations, with all inputs sourced and reproducible). Outputs: open datasets, an open bilingual dashboard, and a conference paper.
A weekly, openly published monitoring of public political messaging and large-language-model behavior, built and maintained by the research team as a public record for journalists and researchers.
Contact โ Civic Data Research Program, JNC TV, USA ยท tcgyver@jnctv.com
JNC TV๋ CA ์บ๋ฆฌํฌ๋์์ฃผ์ ๊ธฐ๋ฐ์ ๋ ๋น์๋ฆฌ ํ์ธ ์๋ฏผ ๋ฏธ๋์ด ๊ธฐ๊ด์ ๋๋ค. JNC TV๋ ์ปค๋ฎค๋ํฐ ์ ๋๋ฆฌ์ฆ๊ณผ ํจ๊ป, ๊ตญ๊ฐ ๋ณด๊ฑด์กฐ์ฌ ์๋ฃ์ ์ ๊ฑฐ ๊ธฐ๋ก ๋ฑ ๊ณต์ ๊ณต๊ณต๊ธฐ๋ก์ ๊ฒ์ฆ ๊ฐ๋ฅํ๊ณ ์ฌํ ๊ฐ๋ฅํ ํยท์ ์ด์ค์ธ์ด ๊ณต๊ณต ์๋ฃ๋ก ๋ง๋๋ ๋น์์ ์๋ฏผ ๋ฐ์ดํฐ ์ฐ๊ตฌ ํ๋ก๊ทธ๋จ์ ์ด์ํฉ๋๋ค.
์ด ํ๋ก๊ทธ๋จ์ ๋ชจ๋ ์ฐ๊ตฌ๋ ๋น์์ ์ ์ผ๋ก ์ํ๋ฉ๋๋ค. ๋ฐ์ดํฐ์ ์ฝ๋, ์ฐ๊ตฌ ๊ฒฐ๊ณผ๋ ๊ณต๊ฐ๋ฅผ ์์น์ผ๋ก ํ๋ฉฐ, ๋ฐํ๋๋ ๋ชจ๋ ์์น๋ ์๋ณธ ๊ธฐ๋ก๊ณผ ๋์กฐํด ๋ ๋ฆฝ์ ์ผ๋ก ๊ฒ์ฆํ ๋ค ๊ณต๊ฐํฉ๋๋ค.
์ฐ๊ตฌ์ฑ ์์ / ๊ณผํ ์ด๊ด
์ ํฌ๊ฒฝ ๋ฐ์ฌ๋ JNC TV ์๋ฏผ ๋ฐ์ดํฐ ์ฐ๊ตฌ ํ๋ก๊ทธ๋จ์ ๋ชจ๋ ์ฐ๊ตฌ์ ์ค๊ณยท๋ถ์ยท๊ฒ์ฆ์ ์ด๊ดํฉ๋๋ค. ๋ฏธ๊ตญ ์ง๋ณํต์ ์๋ฐฉ์ผํฐ(CDC)์์ ๋ณด๊ฑด ๋ถ์์ ์ํํ ์กฐ์ฌํต๊ณํ์์ด์ ์์ ๋ฐ์ดํฐ ์ฌ์ด์ธํฐ์คํธ๋ก, ๋ณตํฉํ๋ณธ ์ค๊ณ์ ๊ฐ์ค, ์ถ์ธ ๋ถ์, ์ฌํ ๊ฐ๋ฅํ ํต๊ณ ํ์ดํ๋ผ์ธ์ด ์ ๋ฌธ ๋ถ์ผ์ ๋๋ค. ํ์ฌ ๋ฏธ๊ตญ ์ฒญ์๋ ์ํํ๋์กฐ์ฌ(YRBS) 2015โ2025 ๋ฏธ์์๋ฃ ์ฌ๋ถ์๊ณผ ๊ณต์ ์ ๊ฑฐ ๊ฐํ ๊ธฐ๋ก์ ํต๊ณ์ ๊ฒ์ฆ์ ์ํํ๊ณ ์์ต๋๋ค.
์ฐ๋ฝ์ฒ: tcgyver@jnctv.com
๋ฆฌ์์น ์์ง๋์ด / ๋ฐ์ดํฐ ํ๋ก๋ํธ ์ด๊ด
ํ์ธ์ฐฌ์ JNC TV ์คํํ๋ก, ์ธํฐ๋ํฐ๋ธ ๋์๋ณด๋์ ์ฃผ๊ฐ ๊ณต๊ณต๋ฐ์ดํฐ ๋ชจ๋ํฐ๋ง ๋ฆฌํฌํธ ๋ฑ ์ฐ๊ตฌ ํ๋ก๊ทธ๋จ์ ๋ฐ์ดํฐ ์ ํ๊ณผ ์ฝํ ์ธ ํ์ดํ๋ผ์ธ์ ๊ตฌ์ถํฉ๋๋ค. ๋ฐด๋๋นํธ๋ํ๊ต ์กธ์ (์ ์นํยท๊ฒฝ์ ํ, ์ปดํจํฐ๊ณผํ ๋ถ์ ๊ณต).
์ฐ๊ตฌ ํ๋ ฅ์
์กํ ๋ฐ์ฌ๋ ๋ฐด๋๋นํธ๋ํ๊ต ์ฐ๊ตฌ ์ฌ์ด์ธํฐ์คํธ๋ก, ํยท๋ฏธ ์ ๊ฑฐ ๋ฐ์ดํฐ ๊ฐ๋ฐฉํ ๋ถ์ ํ๋ซํผ์ธ Electoral Insights Hub๋ฅผ ํฌํจํ ์ ๊ฑฐ ๋ฐ์ดํฐ ์ฐ๊ตฌ์์ ๋ณธ ํ๋ก๊ทธ๋จ๊ณผ ํ๋ ฅํฉ๋๋ค.
CDC ์ฒญ์๋ ์ํํ๋์กฐ์ฌ(YRBS) ๊ณต๊ฐ ๋ฏธ์์๋ฃ๋ฅผ ์ ๋ฉด ์ฌํ ๊ฐ๋ฅํ๊ฒ ์ฌ๋ถ์ํด, 2021๋ ์ดํ ์ฒญ์๋ ์ ์ ๊ฑด๊ฐ๊ณผ ์ํํ๋์ ๊ฐ์ ์ด ์ฑ๊ณผ ์ธ์ข ยท๋ฏผ์กฑ ๋ชจ๋ ์ง๋จ์ ๋๋ฌํ๋์ง๋ฅผ ๊ฒ์ฆํฉ๋๋ค. ์ฑ๊ณผ๋ฌผ: ๋๋ฃ์ฌ์ฌ ๋ ผ๋ฌธ๊ณผ ํยท์ ๊ณต๊ฐ ์์ฝ.
ํ๊ตญ ๊ณต์ ๊ฐํํํฉํ 18,949์ฅ์ ๋์งํธํ์ ํต๊ณ์ ๊ฒ์ฆ, ํฌํ๊ตฌ ๋จ์ ๋ถ์ ์ฐ๊ณ, ๊ทธ๋ฆฌ๊ณ 2026๋ ๋ฏธ๊ตญ ์ค๊ฐ์ ๊ฑฐ๋ฅผ ์ํ ๋ณด์ ๋ ๋ชฌํ ์นด๋ฅผ๋ก ์์ธก ๋ชจํ(์๊ด์ค์ฐจ ์๋ฎฌ๋ ์ด์ 40,000ํ, ๋ชจ๋ ์ ๋ ฅ๊ฐ ์ถ์ฒ ๋ช ์ยท์ฌํ ๊ฐ๋ฅ). ์ฑ๊ณผ๋ฌผ: ๊ณต๊ฐ ๋ฐ์ดํฐ์ , ํยท์ ๊ณต๊ฐ ๋์๋ณด๋, ํํ ๋ ผ๋ฌธ.
์ฐ๊ตฌํ์ด ๊ตฌ์ถยท์ ์งํ๋ ์ ์น ๋ฉ์์ง์ ๊ฑฐ๋์ธ์ด๋ชจ๋ธ ํ๋์ ๋ํ ์ฃผ๊ฐ ๊ณต๊ฐ ๋ชจ๋ํฐ๋ง์ผ๋ก, ์ธ๋ก ์ธ๊ณผ ์ฐ๊ตฌ์๋ฅผ ์ํ ๊ณต๊ณต ๊ธฐ๋ก์ผ๋ก ๋ฐํํฉ๋๋ค.
์ฐ๋ฝ์ฒ: JNC TV ์๋ฏผ ๋ฐ์ดํฐ ์ฐ๊ตฌ ํ๋ก๊ทธ๋จ, ๋ฏธ๊ตญ ยท tcgyver@jnctv.com
Send me a message or reach out via social links.
Senior Data Scientist | Healthcare | NLP | MLOps
hkchun18@gmail.com
Nearly two decades building and validating predictive, causal and detection models on large real-world data. Azure/Databricks, Snowflake, AWS, NLP, MLOps and agentic AI.
Data Scientist โ OhMyCode LLC (2025โPresent)
Statistician / Data Scientist โ CDC (2022โ2025)
Data Analyst โ CDC (2019โ2022)
Economist / Health Scientist โ CDC (2008โ2017)
Sc.D., Work Environment Policy โ UMass Lowell
M.A., Economics โ Boston University
Python, PySpark, SQL, R, SAS, XGBoost, HuggingFace, Spark NLP, Azure ML, Databricks, Snowflake, AWS Bedrock, Power BI, MLflow
Azure AI-900, DP-900, AZ-900