Six roles in three cities, each closer to the small business

Mandate

Merge four country products into one platform with a bounded compliance layer for each market, make pricing work for micro businesses and introduce machine-learning features without putting customers' books at risk.

Outcomes

• Merged four country codebases into one shared core with four country compliance layers by the end of 2022.
• Built MyInvois e-invoice submission into every Malaysian plan; about 2.6 million e-invoices a month by June 2026, with monthly churn down from 2.3 to 1.6 per cent.
• Cut abandoned invoice sessions outside metropolitan Indonesia and the Philippines from 38 per cent to 9 per cent with an offline-first app.
• Grew paying businesses from 196,000 at the end of 2021 to 412,000 in June 2026, helped by per-business pricing and the Mikro plan.
• Reduced partner lenders' fraud losses on invoice financing from 0.9 per cent to 0.08 per cent of advanced value.

Decisions in context

I became chief product officer in July 2021, when the co-founder who had led product moved to run engineering as chief technology officer. Rimbunara then had about 196,000 paying businesses and four products that shared a brand but little else. My mandate from our chief executive was to make them one platform without losing what each country needed, and to fix pricing for the smallest businesses before we spent heavily on acquiring them. The first two years were structural: a shared core, four country compliance layers, country product directors with real authority and an offline-first mobile app. From 2024 the work turned to Malaysian e-invoicing, receipt reading and the bookkeeping assistant, and in 2025 to rebuilding trust controls after the invoice-financing fraud. I lead 236 people through nine direct reports, write a product paper for every board meeting and report on trust and safety to the Audit and Risk Committee.

Reason for the move

Current role. Not seeking an executive move; considering one independent directorship with the consent of Rimbunara's board.

Mandate

Launch the Philippine product, rebuild Indonesian onboarding and prepare four separate country products for consolidation.

Outcomes

• Launched payment links in April 2020; they collected about US$310 million for small businesses in their first twelve months.
• Launched in the Philippines in September 2020 and reached 21,000 paying businesses by June 2021.
• Cut Indonesian onboarding from fourteen steps to six, raising completed sign-ups by 37 per cent.

Decisions in context

I joined Rimbunara in December 2019 to run product for its Malaysian and Indonesian businesses and to open the Philippines. Within three months the pandemic had closed shops across all four markets, and the most urgent request from customers was a way to be paid without a counter. We shipped payment links that a seller could send through a chat app in five weeks, and in their first year they collected about US$310 million for small businesses. The Philippine launch went ahead in September 2020, run remotely with a Manila team I met in person only the following year. I also rebuilt Indonesian onboarding, which had asked for fourteen pieces of information before a business could issue its first invoice. Working across four separate codebases convinced me, and then our chief executive, that the company needed one platform with clearly separated country rules.

Reason for the move

Promoted to chief product officer when the co-founder who had led product became chief technology officer.

Mandate

Turn a document-checking tool into a verification platform that institutions could trust with remote onboarding.

Outcomes

• Launched face matching with liveness checks, used in about nine million verifications a year by 2019.
• Cut false rejections of genuine applicants holding older identity cards from 14 per cent to 4 per cent.
• Delivered verification for a state government's online business-licence renewals, used by about 60,000 businesses a year.
• Kept every customer contract through the 2018 acquisition and integration.

Decisions in context

Veritrau sold identity verification to institutions that had to decide, remotely, whether an applicant was who they claimed to be. When I joined in November 2015 the product checked identity documents; by 2019 it matched faces to documents with liveness checks and handled about nine million verifications a year for banks, insurers, telecommunications companies and a state government's business-licensing portal. The work taught me how much harm a false rejection does. Older identity cards with faded photographs were causing 14 per cent of genuine applicants holding them to be turned away, most of them older people, and we spent a year with our data scientists and a panel of customers' branch staff bringing that to 4 per cent. In 2018 a regional telecommunications group acquired the company. I stayed through the integration, but as our roadmap turned towards the group's own needs I wanted to build products that small businesses used directly.

Reason for the move

Wanted to build for small businesses directly after the roadmap shifted towards the acquiring group's own needs.

Mandate

Add the payment methods Southeast Asian shoppers actually used and make merchant onboarding faster without letting fraud in.

Outcomes

• Added fourteen local payment methods across six markets, including bank transfers, cash payments at convenience-store counters and early e-wallets.
• Cut merchant onboarding from nine days to two.
• Reduced chargebacks from 0.62 per cent to 0.31 per cent of card volume.

Decisions in context

I moved to Singapore in August 2011 to join Quilmara, an online payment gateway for e-commerce merchants in six Southeast Asian markets. Shoppers in the region paid in ways that card-centred products ignored: by bank transfer, in cash at convenience-store counters with a reference number, and later through e-wallets. Each method failed differently, and each attracted its own kind of fraud. As senior product manager and, from 2013, director of product for merchant onboarding and risk, I added fourteen local methods, cut onboarding for new merchants from nine days to two and halved chargebacks on card volume. Those results came from making onboarding stricter in a few places and simpler in many. I completed an MBA part-time at the Selatara School of Management during these years, graduating in 2014. By 2015 I was convinced that the decisive moment for fraud was onboarding, and I wanted to work on identity itself.

Reason for the move

Saw that most merchant fraud began with an identity that should not have been accepted, and moved to work on identity verification.

Mandate

Make SMS and feature-phone transactions reliable for kiosk owners on weak mobile networks.

Outcomes

• Redesigned SMS commands so that a repeated message could not charge twice; failed transactions fell from 11 per cent to 3 per cent.
• Introduced a float balance message after every sale, cutting support calls per kiosk by 40 per cent.
• Launched a feature-phone application used by 12,000 of the network's 38,000 active kiosks by 2011.

Decisions in context

I moved to Jakarta in March 2007 to join Kiosnusa as a product analyst, the only non-Indonesian on a product team of four. The company supplied airtime, prepaid electricity tokens and bill payments to small kiosks that sold them on to their neighbours, and the kiosks ordered through SMS commands typed on basic phones. My first months were spent in kiosks in Bekasi, Cirebon and Makassar, watching transactions fail. A weak signal could deliver a command twice or not at all, and the kiosk's float was debited either way. As product manager from 2009 I rewrote the command syntax around a reference that made repeats harmless, added a balance message after every sale and launched a small application for feature phones. Failed transactions fell from 11 per cent to 3 per cent. I left in 2011 to work on payments across the region rather than in one market.

Reason for the move

Wanted to work on online payments across several countries and moved to Singapore.

Mandate

Build and maintain till software and, from 2005, lead field support for stores in the Klang Valley and Perak.

Outcomes

• Built an offline receipt queue that kept tills selling when a store's link to head office failed.
• Wrote the first support handbook for store staff, in Malay, English and Chinese.
• Cut repeat field visits per store from 2.1 to 1.3 a quarter.

Decisions in context

My first job after graduating was as a software engineer at Damarlis Systems, whose point-of-sale software ran in about 1,900 minimarkets and pharmacies in Peninsular Malaysia. I wrote till software for two years and volunteered for field support whenever a store in the Klang Valley or Perak reported a fault, because I wanted to see the software where it was used. In 2005 I became field support lead, with a small team of engineers. The most useful thing we built was an offline receipt queue, which let a till keep selling when the store's connection to head office dropped and sent the records on later. I also wrote the first support handbook for store staff, in Malay, English and Chinese. Shopkeepers taught me that a till they understood was worth more to them than any report. In 2007 I accepted an offer in Jakarta to work on mobile transactions for small sellers.

Reason for the move

Wanted to work on mobile transactions for small sellers and accepted a role in Jakarta.