Independent software product & engineering studio
Ideas into durable systems.
Build. Verify. Operate.
We design and build software for real problems — from products and operational tools to applied AI and experimental systems.
Capabilities overview
What we can build
Local-first Products
Browser extensions and desktop tools with clear, testable on-device data boundaries. User data stays on the machine unless the user chooses otherwise.
Review Scout · Volume Booster
Read-only Systems Tooling
Deterministic forensic tools that trace evidence to file and byte offsets without modifying the stores under investigation.
MQ Watcher
Local AI & Research Runtimes
High-throughput inference recipes on consumer GPUs and continuity-gated generation pipelines where evidence precedes canon.
Qwen 5090 Recipe · Longform Engine
Selected work
What we've built
01Review Scout
PreviewA local-first Chrome extension for Amazon review analysis — on-page parsing, multi-language comparison, and evidence-backed ranking, with all data stored in extension-origin IndexedDB and no remote analysis backend.
2026
02Abyss Survivor
Release CandidateA mobile-first space survival action roguelike built on pure HTML5 Canvas and procedural Web Audio — engineered for instant startup and fluid 60fps performance without commercial engine bloat.
2026
03MQ Watcher
PreviewA read-only evidence explorer for Apache ActiveMQ Classic KahaDB stores — parse journal files locally, correlate message, ACK, subscription, and transaction evidence down to file and byte offset, without starting a broker or touching the store.
2026
Engineering method
How we turn a problem into a system that works in practice
Constraints & Architecture
We define operational boundaries before writing code. Architecture decisions (ADRs) are recorded with their trade-offs.
Evidence-backed Outcomes
Every claim carries its measurement conditions. We test in real browsers and device viewports with hash-verified fixtures.
Observability & Runbooks
Launch is day one, not the end. We deliver explicit runbooks, rollback plans, and interpretative boundaries for every system.
Active explorations
Testing ideas before they become products
Technical Support Specialist
Distilling domain-specific diagnostic reasoning for enterprise middleware into compact models, with evidence checks and a fail-closed gate for destructive commands.
※ An active research hypothesis, tested separately from production environments.
Longform Continuity Engine
Cross-episode narrative continuity testing where world state is authoritative, evidence comes before canon, and failed checks block commits.
※ An active research hypothesis, tested separately from production environments.
Studio identity
We build products and systems that hold up in real use.
GYUFORGE is an independent software product and engineering studio. We define the problem, build the system, verify its behavior, and keep refining it until it can operate under real constraints.
Next step
Working on a real problem?
Tell us about your project, what you need, and any constraints. We’ll review the details and discuss the next steps.