Spacewalk
Utilizing ML Algorithms in the Real Estate Industry

Sunghyeon Cho, the Founder and CEO of Spacewalk, has always had a particular interest in architecture, real estate, and technology. While studying architecture and computer engineering at Seoul National University, he worked on architectural design challenges and came up with the idea that computer design programs could automate a significant portion of the architectural process. This was when the first concept of architectural design technology came into his mind.

In 2013, he, along with his team, founded ‘Boundless’ architecting office, and over three years, he conducted the design of 20 buildings and 300 design strategies for these buildings. In 2016, Mr. Cho took a technical team from the ‘Boundless’ architecting office to start Spacewalk—a deep tech startup—and began to optimize land use through AI architectural design technology. Starting with developing the “automatic program for small-scale housing reconstruction on streets” for Seoul Housing & Urban Corporation (SH) in the same year, the company launched its flagship product equipped with Deep Reinforcement Learning AI technology in 2018.

Today, Spacewalk offers two leading technologies, AI architectural design and AVM (Automated Valuation Model), through its flagship product, Landbook. AI architectural design technology calculates the value that can be added through a construction project. It enables the company to generate the most profitable architectural plan abiding by more than 20 construction-related laws and regulations. “We are developing machine learning technology based on the expertise and intuition of architects to solve the difficult problem of finding the optimal solution among more than 10^210 numbers of cases,” says Daniel Lee, the CTO of Spacewalk.

AVM technology also measures the transaction value of a property. It enables Spacewalk to accurately predict the future price using past transaction records and different property-related information that may affect the future value. “The goal of the whole technology is the “highest and best use” of the land. It has been our mission since we started our business as an AI architectural design technology company,” states Jeff Choi, CPO of Spacewalk. However, this objective cannot be achieved without the prediction of the market value of the property. This is where Spacewalk’s AVM technology contributes. AVM helps the AI architectural engine to select the most profitable scenario among millions of generated designs. With AVM and architectural technology put together, the future value of the land after each development project can be calculated.

Started as a technology-based solution provider, Spacewalk is now evolving into a platform company that revolutionizes how both the supply and demand sides find each other on the Korean property market. Landbook acts as an AI expert for people who are trying to buy and sell unstandardized properties. “Through the platform, we are connecting the buyer and seller of the unstandardized properties,” mentions Henna Lim, Head of business of Spacewalk. Further, at Spacewalk, domain knowledge experts such as architects and real estate investors are involved in every step of technology projects and provide their experiences and insights acquired in their field to technology experts. “Almost 70 percent of our team members are engineers developing AI technology with the combination of machine learning, data science, and architectural algorithms,” states Eddie Tae, finance officer of Spacewalk. As a result, the company’s technology inevitably bears high applicability in the market. It is truly a no brainer that Government-owned corporations such as Korea Land & Housing Corporation (LH), Seoul Housing & Communities Corporation (SH), Gyeonggi Housing & Urban Development Corporation (GH), and Incheon Housing and City Development Corporation (iH) are using Spacewalk’s product to get their job done.

Further, Spacewalk’s AgTech team participated in the Autonomous Greenhouse International Challenge (AGIC) 2019 as a member of the Korean national team and designed algorithms to grow tomatoes to the maximum value. Spacewalk has proven its technological viability by winning 2nd place in the preliminary round and third in the final round among 21 teams globally. “We are trying to improve the efficiency of agricultural land which takes up 20 percent of South Korea’s land. We are applying deep reinforcement learning technology in developing automated planting algorithms,” concludes Mr. Cho.