Hi, I'm Hyoyeon Lee. I build end-to-end data & ML systems for semiconductor and finance. My interests include decision-making, optimization, robust/trustworthy ML, and quantitative research.
Work Experience
- AI model optimization at large-scale manufacturing sites. On real DRAM mass-production lines, deployed ML optimization frameworks on high-volume test streams and achieved record-high yields for new products. Engineered Bit Line Sense Amplifier (BLSA) optimization with Python and C/C++; built a cloud edge-computing–based platform that performs real-time optimization and automation across multi-device testing environments. Afterward, I directly set up the ML optimization system for all incoming DRAM product requests.
- Big-data & platform engineering. Designed and operated a TB-scale parallel processing framework using Dask, PyArrow, and AWS Glue to handle production logs. Fully automated anomaly detection via Airflow pipelines. Authored seven reusable automation tools that reduced repetitive tasks by ~90%, with results validated by team and division awards.
- Independent R&D (Algorithms · Causal/RL). Performed defect-cause inference using ML data analytics, hybrid ML tuning for process parameters, and robust reinforcement learning. Results include an IEEM-accepted paper and presentation, two patents, and CEO/Best Paper awards.
- Leadership · Communication · Education. Served as Culture Agent for a 50-member team and taught internal ML seminars (won Best Seminar Award). Consistently connected seminar/research insights to independent decision-making on new algorithm development projects.
Research Experience
- Statistics Consultant, Korea University Statistics Consulting Center (2019–2021). Analyzed chronic kidney disease progression using survival models on clinical time-series data. Identified early dementia biomarkers via structural equation modeling and cluster analysis. Evaluated impact of teaching methods through factor analysis and regression in child education study.
- Research Assistant, National Statistical Research Center (2018–2019). Performed Bayesian change-point survival analysis to evaluate vaccine effectiveness and durability. Developed statistical learning methodologies for personalized medical treatment recommendations. Conducted semi-parametric survival analysis for dependent censoring and dynamic treatment regimes.
(More details on CV.)
Selected Projects
Project details and demos are organized on the Projects page.
DRAM ML Optimization
Production-grade optimization + automation pipeline for high-volume test streams.
Company-internal project; details are omitted due to confidentiality.
TB-scale Data Platform
Parallel processing + orchestration for production logs; anomaly detection workflows.
Company-internal project; details are omitted due to confidentiality.
☀️ Festory - Festival Travel Website
AI-powered festival recommendation platform with Google Maps/Calendar integration and real-time weather.
🧬 Alpha-Helix - Prompt-to-Portfolio Quant System
End-to-end quantitative research system: ingestion → signals → evaluation dashboards.