Observed arrival · 2026-09-22
Chang Hyun Woo’s Measured AI Experiments
A Seoul-based AI engineer’s portfolio of LLM agents, live demos, benchmarks, and postmortems.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- AI builders, agent researchers, and curious technical readers
- Worth noticing
- The homepage exposes held-out test results, latency changes, and a benchmark where two of four rows were lost.
Field notes
The portfolio treats measurement as part of the work rather than a marketing afterthought, showing held-out cases, paired tests, latency changes, and corrections to its own benchmark. Lumina’s described pipeline includes speaker-specific memory retrieval, a local voice-addressing gate, streamed speech, and local audio playback for lip synchronization. Related sections identify OpenVault by its MCP-tool and test counts, while the site also links to public source, recordings, and playable experiments.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
◉HumanPersonal, local, civic, or handmade
One card from the complete issue