6954

한국어로 읽기 →

About

Seongkyoun Yoo

Engineer · Builder · Productization Practitioner

I build systems, measure what actually happens, and turn the useful ones into products and public evidence.

GitHubProjectsContact

From low-level systems to AI products

I started with low-level software — embedded Linux, BSPs, device drivers, NAND flash and communication stacks.

Over more than two decades the work moved upward: devices, distributed systems, remote management, AI inference, and eventually productization.

The common thread has stayed the same: taking something uncertain and making it operational.

Build · Measure · Productize

Build

Embedded systems · Linux · Edge AI · distributed systems

Not ideas about systems — systems that run.

Measure

Benchmarking · instrumentation · failure analysis · operational limits

Not stopping at "it works" — how much, and where it breaks.

Productize

Packaging · positioning · deployment · go-to-market

Not stopping at a demo — a form that can leave the building.

Selected work

Open source · Edge NPU Cluster

NPUDure

Open-source distributed inference runtime built on three RK3576 NPUs. 421 hardware measurements, 3.00× measured 1→3 node scale-out — and the findings that almost stopped it.

View projectEvidence

Product · live

AI Arcade

A browser-based collection of small AI experiences designed around direct interaction rather than passive demos.

Open

Applied AI research · in progress

Vending AI

An empirical AI project exploring when adding more structure improves — or fails to improve — operational outcomes.

Results will be published here when the study closes.

How I work

Measure before optimizing.

If the bottleneck has not been measured, optimization is speculation.

Publish the limitations.

A result without its conditions is usually less useful than it looks.

Ship before polishing forever.

A project sitting on a hard drive has no external value.

Prefer evidence over narrative.

When measurements contradict the plan, the plan loses.

Background

I have spent more than two decades building software across embedded systems, Linux, device communication, remote management, printing infrastructure and AI systems.

My early work was close to the hardware — BSPs, drivers, communication stacks and storage. Later work expanded into distributed systems, product development, open source, AI inference and technical commercialization.

I am a doctoral student in the Department of Venture and Small Business at Soongsil University.

Research & writing

My current interests sit around technology adoption, productization, operational readiness, and the gap between technical capability and organizational results — the distance between building something and getting it used.

Published papers, talks and datasets will be listed here as they appear.

Let's talk

If you are working on Edge AI, embedded systems, distributed inference, or turning technical prototypes into deployable products, feel free to reach out.

GitHubDiscussions