In March 2025, I traveled to San Jose for NVIDIA GTC 2025. I arrived mainly hoping to see new technology firsthand and meet people working on it. After several days of sessions and conversations, however, one point had become much clearer: change in AI was no longer confined to model performance. It was spreading across infrastructure, products, and the physical world.
A smart contract is often introduced as “a contract that executes automatically when its conditions are met.” It is an intuitive definition, but after building and operating these systems, I find that it leaves out something essential. A smart contract does not understand the real world or interpret what its parties intended. It is better understood as an execution rule applied consistently by a blockchain network.
Working in blockchain, I am often asked what the future of Web3 looks like. I used to answer with familiar ideas: decentralization, data ownership, and token economies. None of that is wrong. But after building and operating products and mainnets, I have come to think a different question matters more: not what Web3 promises, but what it can genuinely solve better than existing systems.
At GDAC Exchange, I built and upgraded systems supporting anti-money-laundering operations. The project connected customer, trading, deposit, and withdrawal data to an AML solution and integrated KYC, risk assessment, watch-list filtering, and fraud-detection functions into one exchange workflow.
To prepare GDAC Exchange for Korea’s Virtual Asset User Protection Act, I helped develop a market-surveillance system based on the rules and guidance of the Financial Supervisory Service and Financial Services Commission. The work translated regulatory requirements into a system that detected suspicious patterns in near real time and connected each result to reproducible evidence and reporting.
GDAC DeFi brought decentralized-finance functionality into the experience of a centralized exchange. I integrated external market data, contributed to smart-contract development, and analyzed behavior after launch to test our product assumptions. The work connected technical implementation and product judgment because CeFi operations and on-chain execution had to function as one service.