Hackathons produce scam simulation tool for education, platform identifying birds based on calls

SINGAPORE – A scam simulation where users role-play falling prey and losing money to scammers. A web-based game that teaches users responsible pet ownership. A platform that helps to identify local bird species based on their calls.

These are among projects that were created during two hackathons in the last two years: the Hack For Public Good (HFPG) in 2025 and Build for Good (BFG) Environment in 2024.

Organised by Open Government Products (OGP), an independent division of the Government Technology Agency (GovTech), HFPG is a month-long hackathon held every January since 2020 that encourages its employees to build solutions that tackle public issues.

The BFG Environment, organised by OGP in partnership with the SG Eco Fund, is open to all Singapore residents aged 18 and above. Here, five winning teams will take part in an eight-week accelerator programme to develop their innovations, with up to $100,000 in funding, mentorship by industry experts, and product workshops.

Innovations that have emerged from these hackathons include ScamShield, the anti-scam app that helps people filter scam messages and calls, and RemediSG, which collects and redistributes unused medication.

The Straits Times sat down with three hackathon teams to find out what societal issues they are trying to solve, how their creations have evolved since the hackathons, and what plans they have.

Many people have probably seen countless anti-scam advertisements in public spaces and on social media, not to mention read about scam victims in multiple news reports. But have they ever assessed their own abilities to identify a scam?

They can go ahead and try out Unpacked, an immersive scam simulation built by a group of five OGP employees, which allows users to experience first-hand what it is like to interact with scammers posing as government officials.

Created during HFPG 2025, the educational tool complements ScamShield, the anti-scam app that helps people filter scam messages and calls.

“The way citizens consume scam education, such as via advertisements in MRTs and on social media, is very passive,” said software engineer Rayner Loh, one of the team members.

“These traditional methods are not particularly effective in reducing scams and making the public more aware or able to protect themselves… We want to show things like how psychological manipulation happens in real time.”

Accessed via a mobile device on

, the multi-sensory experience starts with a call from an unknown number.

The “caller”, which is actually a voice created digitally, would then identify herself as a DBS Bank employee. The user is told that there has been unusual activity on his or her bank account, and that the bank suspects it is the result of identity theft.

The call is then transferred to the “police”, and the conversation continues on a messaging platform that resembles WhatsApp. Throughout the experience, the user can respond based on choices provided.

The simulation ends when the user simulates transferring his or her life savings to an unknown bank account.

The manner in which scammers speak, type, and the communication channels used are specifically crafted based on information from advisories that the police have put out, and interviews with officers and victims of government official impersonation scams (Gois).

“Some of the insights were that the scammers would sound local, and they would be wearing a police uniform during the video call,” said Mr Loh.

“For Gois, it’s also never on a single platform – they’ll pass the victims around different platforms such as on phone calls and WhatsApp.”

He added that while the team is working to simulate other scam variants in the future, it chose to focus on Gois first due to its high financial impact.

There were 1,504 Gois cases reported in 2024, and an average of $100,622 was lost per case, according to the annual scam figures released by the police in February.

This is the highest amount of money lost per case on average in 2024, compared with other scam types such as investment or social media impersonation scams.

Unpacked has hosted around 7,400 unique visitors since February when it was displayed during the HFPG demonstration day.

The team is working on making improvements such as making the simulation available in Chinese, Malay, and Tamil.

There are no plans now to integrate Unpacked as a feature in ScamShield’s app, but the team continues to share about the simulation through word-of-mouth and online, said Mr Loh.

“However, if there are organisations that are interested in using Unpacked for their own anti-scam education and training purposes, we welcome the public to use it,” he added.

A record number of animal welfare and cruelty cases in Singapore prompted the creation of Tamagotcha – a game inspired by the popular 1990s pocket-size gadget on which players care for a virtual pet.

Created by a team of five during HFPG 2025, Tamagotcha uses gameplay to teach existing and potential pet owners the responsibilities of caring for a furry companion.

Earlier in January, the Society for the Prevention of Cruelty to Animals (SPCA) reported that the number of cases of

, with a total of 2,190 animals falling victim in 2024.

But it was not just statistics that inspired the team. Being pet owners themselves, they knew that new owners can be caught off-guard by the needs of their pets.

“As a first-time cat ‘pawrent’ some years ago, I was really surprised how different their needs are from that of dogs that I had prior experience with – such as their innate need to scratch, play, and explore,” said Ms Nitya Menon, lead policy and transformation specialist at OGP.

“With so many unanticipated scenarios that can make a new pet owner feel ill-prepared, we wanted to focus on upstream education to help new adopters make informed decisions.”

Currently, players can choose between only a cat and a dog, although some pieces of advice given throughout the game are applicable to all pet owners.

To collect information for the game while it was in development, the team surveyed 129 people based in Singapore including pet owners, community animal caretakers and rescue operators. The prototype was also tested by SPCA, and Metta Cats And Dogs Sanctuary.

Some key insights were that healthcare, nutrition, training and costs were the most common areas where pet owners felt they lacked enough information.

The game, which is currently a prototype,

. Users can choose to play with Snowy, a cuddly and affectionate dog, or Nala, a feisty and fearless cat.

They start with nine hearts and nine coins. The goal is to maintain good care of the pet, which would result in a decrease in coins (expenditure on pet care), but an increase in hearts (the pet is happy).

In one instance, the player is confronted with a situation in which Nala pounces and tries to bite the owner’s feet. The player can either choose to scold, or set aside playtime and provide suitable toys for Nala.

“Another area that often catches new owners by surprise is the cost of caring for a pet,” said Mr Goh Kyi Yeung, team member and assistant director at the Ministry of Digital Development and Information.

Basic needs such as cost of food and toys, licensing, grooming, and vet visits can add up quickly, he added. “The game includes scenarios around these practical aspects too, so players can better appreciate the financial commitment involved.”

Moving forward, the team hopes to partner with agencies and animal welfare groups to integrate the game into adoption processes, licensing requirements, or broader public awareness initiatives.

Said Ms Menon: “We welcome pet owners and shelters to try Tamagotcha out and give us their feedback. This will help us decide on next steps, including whether OGP should invest additional resources to expand the game’s role as a public education tool.”

“Our longer-term vision is for Tamagotcha to inspire the reimagination of other learning and training platforms by the Government, from Driver Improvement Point System (DIPS) retraining to election duty,” said Mr Goh, referring to the course meant to correct driving behaviour of errant motorists who have accumulated demerit points.

To make the process of identifying birds easier and less time-consuming when an area is being assessed for environmental risk during development, a group of engineers and developers came together to create an online platform that can identify species of birds via audio recordings.

Koel was created by a team of five during the BFG Environment hackathon in 2024, and the goal is to make environmental impact assessments (EIAs) less challenging to conduct.

Companies and agencies such as the National Parks Board (NParks) commission independent consultants to carry out EIAs, which evaluate the potential impact a new development can have on the environment.

Currently, identifying animals in an area either requires someone to count them manually, or via a camera trap that can detect animals when there is movement. But these methods are not very effective when it comes to birds, said team member and engineer Ervin Cheng.

“Birds are often in trees, and cameras are not pointed upwards and have a limited view,” said Mr Cheng. “Audio recording has a lot of benefits, such as being omnidirectional, and having a range of a few hundred metres.”

At the start of Koel’s development, the team used open source models mostly developed in the US and Europe, which can identify around 900 species of birds that are primarily from these regions.

In order to beef up the platform’s ability to identify birds found in the South-east Asian region, they reached out to Mandai Nature – the conservation arm of Mandai Wildlife Group – which allowed them to record the calls of birds such as the straw-headed bulbul and some hornbills from within its aviaries.

The straw-headed bulbul is a critically endangered bird species, but Singapore is a stronghold for the songbird, with a third of its world population found locally.

The team also chose to record the calls of five other bird species in Mandai Wildlife Reserve, as these birds face potential threats to their natural habitats.

“Our focus currently is to bring up the accuracy of our model, to be able to better identify these species,” said fellow team member and engineer Ritwik Jha.

After users upload their audio recordings, they can also give feedback on whether Koel’s identification was correct – a process the team said will help to refine the system’s accuracy.

Singapore is home to around 400 species of birds, of which Koel is able to detect around 30 to 40 species at the moment. The team is working to increase this number over time.

Koel began working with NParks in mid-2024, and the agency has submitted audio recordings for identification, said Mr Cheng. The main insight that can be derived from analysing audio recordings taken over a period of time is the presence and absence of species, and the general frequency of how often a species was heard calling in an area.

Aside from birds, the platform is also able to identify reptiles and amphibians such as frogs, even though it is not able to distinguish between species currently.

“Mammals tend to not create as much sound, so there aren’t as many audio recordings available as open source models,” said Mr Ritwik. “But animals like frogs are a very good indicator of how healthy the biodiversity and habitat is.”

For now, Koel accepts audio recordings only from NParks and not the community, as there are concerns about bad actors possibly using it for the wrong reasons.

“If the platform is made available to everyone, there could be potential for people interested in poaching or capturing endangered animals to also use Koel,” said Mr Ritwik.

“Perhaps at some point, if we develop the ability to segregate endangered species and keep that data secure, then we might open it up to the public.”