Building in public

Sukmin Yoon

1% better, every build.

Data engineer & AI builder. I've shipped ETL pipelines at scale for sports betting. Now I build AI-powered products in public — and get 1% better at both with every commit.

See the projects
Open to Work — Data Engineering / AI Engineering Roles
About

I built pipelines for ESPN Bet. Now I build products with AI.

Data engineer with a decade of experience building production ETL infrastructure — sports betting, insurance, fintech. My last role was architecting Airflow pipelines and data observability systems for TheScore, partnered with ESPN Bet.

Now I'm doing both: looking for my next DE role while shipping AI-powered products in public. Every project here is a real stack, a real problem, a real commit. That's what 1% better looks like.

TheScore × ESPN Bet

Senior Data Engineer

2023 – 2025

Avesis (via QAC)

Technical Data Lead

2021 – 2023

Jewelers Mutual

Data Engineer

2020 – 2021

VRBO · Expedia

Data Migration Analyst

2016 – 2018

Pipeline Engineering

Built and maintained 15+ Airflow DAGs handling millions of nightly betting transactions at TheScore (ESPN Bet). BigQuery, AWS Redshift, SOX-compliant transformation logic, automated observability frameworks — production-grade, not demo-grade.

AI-Powered Builder

Full-stack from FastAPI to Next.js. Using LLMs and modern AI tooling to ship real products — poker analytics, sports data platforms, API validation frameworks. Every project on this page is live code, not a tutorial clone.

Reliability by Design

A decade of finding data failures before they hit production means I engineer reliability into pipelines from the start. Fail-fast DQ gates, schema drift detection, regression suites that test the tests. Pipelines that don't page you at 2am.

Projects

What I'm building — vote for what's next.

Click “I'm interested” on anything that resonates. It tells me where to focus.

Building

onepercentbetter.poker

GTO Defends. We Exploit.

Full-stack poker analytics platform. Parses GGPoker hand histories, quantifies opponent GTO deviations, and surfaces bb/100 exploit edges by position. The backend analytical engine is live — UI is next.

PokerAnalyticsFastAPINext.js
Building

ActionKeeper

DocuSign for poker staking.

Mobile-first staking agreement app. Players and backers draft terms, negotiate turn-by-turn, and lock in a tamper-evident PDF receipt — cryptographically hashed, QR-verifiable. Never holds funds. Revenue on contract generation via Stripe.

PokerFintechFastAPINext.jsStripe
Idea

StackVision

Point your camera. Know your stack.

Augmented reality poker chip counter. Overlay your phone camera on any chip stack and get an instant count — no manual math, no mistakes under pressure.

PokerARMobileComputer Vision
Live

Blue Jays Moneyball

Production ETL meets sabermetrics. If the data's bad, the pipeline dies.

End-to-end Data Lakehouse on Apache Airflow + PostgreSQL. Ingests MLB salary vs. performance data, runs a fail-fast DQ gate that hard-blocks bad payloads, and a regression suite that verifies the guardrails themselves — the same architecture pattern I used at TheScore.

Data EngineeringAirflowPostgreSQLETLPython
Live

TwelveLabs API Validator

JSON-driven. CI-ready. Reliability-first SDK framework.

Modular Python framework for validating TwelveLabs' multimodal video search SDK. Decouples test logic from data via JSON configs, ships a custom GUI batch runner, and covers i18n edge cases (Korean, Japanese, Arabic). Built to the same standard as a production observability tool.

PythonpytestAI APIsSDK EngineeringAutomation