JungChan LEE
TWO PRACTICES · ONE OPERATOR
I work at the intersection of professional audio production and ML systems engineering. End-to-end K-Pop instrumentals on one side; production ML pipelines on the other. Both practices, the same operator — pick a face.
FACE 01 · AUDIO
K-Pop Producer
K-Pop Producer
& Audio Engineer
Solo K-Pop songwriter, beat-maker, and mixing & mastering engineer. End-to-end in Cubase Pro and FL Studio, no external engineer.
KEY METRIC
−10.5
LUFS · +6.0 LU range
Master delivery on a recent release
ENTER AUDIO ↗
FACE 02 · ENGINEER
ML Systems
ML Systems
Engineer
Self-taught Python pipelines for production ML training and evaluation. LoRA fine-tuning, parameter sweeps, identity benchmarks, real-time monitoring — built solo from data prep to model export.
KEY METRIC
0.6050
ArcFace · vs 0.5090 baseline
AZIT LoRA on FLUX.2 Klein 9B, beat HuggingFace reference
ENTER ENGINEER ↗
WHY BOTH
Audio production taught me to make A/B judgments against commercial
references on every decision — the same instinct that makes
ML evaluation honest. ML systems work taught me to ship a pipeline
end-to-end, document the decisions, and measure against a baseline
instead of a feeling. The two practices reinforce each other; pick
the face that matches what you’re hiring for, or read both.