Software engineering students display computing skills in Didian Hackathon

Front row: Ker (second from left), Lee (third from left), Jackson Siew (third from right)
Back row: Victor Wong (second from left), Choo, Samuel Sim and the Didian team

Bachelor of Science (Honours) Software Engineering student Choo Jia Zheng and Foundation in Science (Sungai Long Campus) student Victor Wong Kim Fung triumphed in Didian HackDay, which was organised on 17 and 18 April 2023 at KL Eco City Strata Office.

The team of duo (Team 1) won a cash prize of RM1,200, Didian HackDay tees, nametags and certificates.

Bachelor of Science (Honours) Software Engineering students Ker Ding Wei and Lee Jia Wei (Team 2); and Samuel Sim Yik Kang and Jackson Siew Kar Soon (Team 3) also participated in the 2-day hackathon. They displayed outstanding performance during the hackathon and were awarded Didian HackDay tees, nametags and certificates respectively.

Didian HackDay was organised by Didian, a company that founded the property marketplace with the same name. The hackathon aimed to incorporate AI (artificial intelligence) technology in developing innovative solutions that make the real estate buying and selling process more seamless for everyone involved. The teams were required to build their own AI-incorporated applications within one and a half days.

Team 1

Choo (centre, left) and Victor Wong (centre, right) with the Didian management

Choo Jia Zheng and Victor Wong Kim Fung’s award-winning project was “FurnitureGPT”. It made use of GPT (Generative pre-trained transformer) to generate furniture in a room. It allows users to take photos of an empty room, select their own furniture and theme, and let the AI generate a fully decorated room.

“People may spend a lot of money just to hire a designer to design a property. In this modern day and age, we thought that ‘Why not just use AI to do this?’ As a result, we came up with the idea of creating an AI designer, which we dubbed ‘FurnitureGPT’. Anyone can use AI for free. Hence, it helps people with little financial resources, who simply want a simple design that suits their preferences. While AI can easily create a design for a room, a designer is still required for a much more detailed and accurate project. Creating a furniture generator bot is quite risky because it requires much more time in model training and the chances of success are extremely low. However, we managed to accomplish it despite having spent so many hours testing the product,” explained Victor Wong regarding “FurnitureGPT”.

When asked about the motivation behind their participation in the hackathon, Victor Wong said, “As I am still a foundation student, I am unable to participate in any closely relevant lecture courses. Yet, I study diligently on my own via the Internet and books. As software engineers, we should frequently challenge ourselves to various projects. Hence, a hackathon is a way for us to become versatile in our skills. During the hackathon, I learned a lot from the other participants by sharing our experiences. Besides, I also learned some knowledge about buying and selling property.”

Choo, on the other hand, said, “I got to know the hackathon mainly through UTAR Information Technology Society. I believe that participating in a hackathon like this can give you a better hands-on experience. You also get to experience a glimpse of what it is actually like to work with people, which is similar to an office environment in a company.”

The proud creators of “FurnitureGPT” were grateful to have each other’s back during the hackathon. Victor Wong enthused, “I am grateful to my teammate, Jia Zheng, for producing excellent work. Without him, we would not have been able to achieve our goal. I would like to thank our mentor as well for assisting us in renting a cloud machine to train the models. He has also been encouraging us throughout the hackathon.”

Meanwhile, Choo enthused, “I feel like what made our project stand out the most was the interactivity. Our product allowed better interaction by letting the judges use their own phones to tinker around with it. Props to my talented and competent teammate, Victor Wong. I am grateful that he supported me throughout the hackathon by creating complex backend codes along with our mentor. This exciting yet challenging journey would not have been possible without their contribution.”

Team 2

Example of a Q&A session between a user and “Agent GPT” regarding Mori Residences

Ker Ding Wei and Lee Jia Wei’s project was “Agent GPT”. Explaining the project, Team 2 said, “Buying or selling a property is extremely time-consuming as many steps must be undergone, including finding a property, and exchanging and signing contracts. It is difficult for property agents to know all the details of the properties. Therefore, to reduce the workload of property agents, we developed an application, ‘Agent GPT’, to help agents in summarising the information of the properties they are handling. The information of the properties is used to train the AI, ChatGPT 3.5. With the help of AI, property agents would not need to memorise or refer to the information of properties. They just need to input the details of properties to the AI and retrieve the AI-generated information of properties based on the agents’ questions. In addition, it can help property agents to generate reasons to buy a particular property, which can be used to generate copywriting materials.”

They added, “‘Agent GPT’ is a very useful tool as it can not only reduce the workload of property agents, but also increase their sales. Besides, customers will be benefited as well because they will be provided accurate information about the properties through property agents. It would help the customers to find their ideal properties.”

“Agent GPT” is in line with the UN Sustainable Development Goal (SDG) 9: Industry, innovation and infrastructure. “It utilises AI technology to streamline property information and provide convenient access to users. Improved infrastructure and innovative solutions can facilitate sustainable development and enhance efficiency in various sectors, including real estate,” explained Team 2.

Team 3

Example of a Q&A chat session with “Property Agent GPT” using three different languages

Team 3 consisted of Jackson Siew Kar Soon and Samuel Sim Yik Kang. Their reasons for joining the hackathon may be different, but their passion for knowledge is the same nonetheless. In his remarks, Jackson Siew said, “I joined the hackathon because my friend is interested in researching and developing a ChatGPT system. Besides, I was looking for properties in the area of Kuala Lumpur as well. Hence, thanks to the hackathon, I was able to learn the complete process of buying a property.”

Meanwhile, Samuel Sim joined the hackathon because he was developing a customised personal virtual character around the time of the hackathon. His endeavour was a success as he had the full system completed and running.

The team’s project was “Property Agent GPT”. Jackson Siew gave an explanation on their project, saying, “Our project, ‘Property Agent GPT’, could represent property agent in providing many information without requiring customers to go to the property showroom. This system could also provide information such as nearby restaurants, the possibility of flood at the location and more.”

“We wish to provide great user experiences to users with our system. Most human agents would choose to direct the users to the showroom. It is time-consuming, therefore, we created ‘Property Agent GPT’ to act as a virtual agent to help property agents in providing information to the users. With properties located in different locations, it is better to filter out the houses that are not suitable for the users. However, to do so, the users must provide proper information such as price to avoid overpriced properties. The virtual agent could even answer questions that some inexperienced agents could not answer. For example, using big data, the virtual agent is able to predict the future price of a property based on historical data, which is more realistic. Moreover, our virtual agent also provides GUI map and 3D live tour for the properties,” elaborated Samuel Sim.

In addition, Team 3 also shared their opinion on the issue of AI as a replacement for human agents. They commented, “The purpose of AI is to assist humans, not replace human jobs. While our personal virtual agent can assist users, it still cannot replace human agents as the AI could not bring users to see the physical property nor do paperwork.”

Wholly owned by UTAR Education Foundation (200201010564(578227-M))         LEGAL STATEMENT   TERM OF USAGE   PRIVACY NOTICE