AI data platform company Unnet on March 19 held a seminar at Conrad Seoul on its Tarantula AI data platform strategy and introduced a financial-sector case of building a Tarantula Lakehouse.

The seminar mainly covered the case of Nextrade (NEXTRADE), a financial data company based on the country’s first alternative trading system (ATS). Nextrade processes average daily trading worth about 15 trillion won, and runs a system where trading can reach 59 trillion won at times. In the process, about 120 GB of data a day and 30 to 40 TB a year continue to accumulate.

According to the company, as data grew rapidly, storage costs kept rising and it was not easy to analyse accumulated data or use it for AI. Because analysis systems were operated separately, management became complicated. There were also difficulties linking to cloud or AI services.

Nextrade introduced the Tarantula Lakehouse to address these issues. The system combines data storage and analysis in a single structure. It enables data to be gathered in one place and immediately analysed or used for AI. It is also connected to the cloud.

Unnet CEO Ju-seop Kim (김주섭) stressed, "Tarantula Lakehouse is a platform verified through actual operations in the financial sector, and a realistic alternative that can improve cost and performance at the same time."

Unnet also unveiled the Tarantula AI VectorDB at the seminar. While conventional databases search for words as they are, a vector DB is characterised by understanding sentence meaning and context to find similar information.

The company stressed that Unnet’s Tarantula AI VectorDB can be stably operated in on-premises environments, making it usable for finance or public institutions where security is important, and that it is designed to deliver fast search performance even in large-scale data environments.