GEM Squared, Inc. · CEO
David Seo
Founder who builds AI output verification infrastructure — hands-on.
With 20+ years of enterprise systems design, I personally design and build the verification layer for AI outputs.
Now
GEM Squared, Inc. — AI Output Verification & Governance
Company
GEM Squared, Inc.
A US Delaware corporation. I founded and operate this startup, which builds the verification layer for AI execution.
Product
TPMN Checker
Compares AI-generated outputs against source materials, task contracts, and governance rules to produce structured audit reports. Identifies unsupported claims and over-extrapolation before they reach production.
Role
CEO & Individual Contributor
I personally execute the full cycle — architecture, implementation, and operations. I am not only a founder, but also the engineer who writes the code and ships the product.
Beyond my own product, I also design and build practical LLM Agent, RAG, and workflow automation systems for teams that need production-oriented AI execution.
Why AI Output Verification?
Unverified AI output is an operational risk.
Next-generation financial systems. Defense simulation. Blockchain platforms. Government-funded R&D. For 20 years I have designed mission-critical domains where verification failures can quickly become operational or business-critical risks.
AI is no exception. If anything, AI produces unverified errors faster and in more plausible-looking forms. Asking the prompt to “please be careful” every time is not engineering.
Without a contract, there is no verification. Without verification, there is no production. The principle is the same whether the system is next-gen banking or AI governance.
TPMN Checker is that principle implemented as a tool for the AI era. The patterns I have seen and lived as an architect are the foundation of what I am building now.
Path
Architect → Founder → AI Governance Builder
Not a chronology — the same problem (system reliability) solved across different eras with different tools.
- 2026 – PresentGEM Squared, Inc.Delaware, USA
CEO & Individual Contributor
Building and operating TPMN Checker, an AI output verification and governance tool. Replacing the fragile prompt-centric paradigm with a system grounded in verifiable execution contracts, scoring, provenance tracking, and human-in-the-loop review checkpoints. Architecture through implementation and operations — all hands-on.
- 2013 – 2022CrowdParti, Ltd.Seoul, Korea
CEO
Principal Investigator on a TIPA (Korea Technology and Information Promotion Agency for SMEs) private-investment-led R&D grant — developed a blockchain-based cross-border content exchange platform (total budget ~KRW 500M). Selected as KOTRA ‘Creative Startup of the Year’ (2015) and KISED ‘Promising Startup.’ Raised USD 400,000 across two seed rounds and participated in the Denmark-government-sponsored Startup Bootcamp.
- 2012 – 2013PwCGangnam, Seoul
Senior IT Consultant
Lead Technical Architect on the ThaiLife next-generation ISP (Information Strategy Planning) project.
- 2006 – 2010BearingPointGangnam, Seoul
IT Consultant
PMO PM on the Solomon Savings Bank next-generation system integration; TA Team Lead and Shared Architecture PM on the NongHyup credit (next-gen banking) system; Team Lead on BearingPoint’s global SOA service localization project.
- 2001 – 2003KCC IT R&D Internal VentureSeoul
R&D Team Lead
Developed a commercial solution (xFrameCMS) on Microsoft C# and delivered SBSi customization. Recognized at the time as one of Korea’s first official Microsoft C# development partners.
- 2000 – 2001KIDA (Korea Institute for Defense Analyses)Seoul
Researcher
TF Lead on the Combat Training Center simulation system development. Contributed to the ammunition information system build-out.
Education: B.S. in Computer Engineering, Incheon National University (1990 – 1994) · ROTC commission
Strengths
Design it. Then build it myself.
Enterprise Architecture
SOA, microservices, and next-generation system design. 20 years of in-the-field experience across finance, insurance, public sector, and large-enterprise domains.
AI Governance
AI output verification, special-purpose AI agents, contract-driven governance, and LLM-based workflow design.
Blockchain · Databases
Blockchain platform construction. Comfortable across SQL (MySQL, PostgreSQL), NoSQL (Redis, MongoDB), and VectorDB (Milvus, Weaviate, Chroma, pgvector).
Hands-on Development
Go, Python, TypeScript, Node.js as primary stack. Heavy Claude Code user; works directly with agent environments including Claude Headless·API, Codex CLI, and Gemini CLI.
Cloud Operations
AWS, GCP, Fly.io, Cloudflare, Vultr — operated directly, not delegated.
Team Leadership · Fundraising
PMO and cross-functional team leadership, seed-round fundraising, and government R&D grant acquisition.
Languages: Korean (Native) · English (business communication; comfortable in global collaboration)
Hackathons
Building, even on short cycles.
In hackathons I personally cover the full cycle — planning, architecture, implementation, pitching, and documentation. Not just ideas, but working software and submission-ready documents. Average contribution: planning 100%, implementation 85%, pitching·documentation 100%.
Together
Projects I want to work on.
Project engagement
- AI governance · AI output verification
- Special-purpose AI agent design
- Enterprise architecture · microservices
- Blockchain platforms · database design and construction
- LLM-based business automation · AI execution verification
Consulting
- US entity incorporation, in practice
- Early-stage tech startup operations
- Government R&D grant planning and acquisition
- Seed-round fundraising preparation