Six ventures.
Every lesson, at cost price.

built

A mobile queue management system — grab a ticket from your phone, watch the line move in real time, show up exactly when it's your turn. I built the Android app, the iOS app, and the web apps for the TV displays and counter kiosks. Alone.

what worked

Frost & Sullivan Product Innovation Award 2015. First cohort of the MaGIC Accelerator. Seed funding from SUN Group Holdings in Singapore. Interviews on BFM, features on TechInAsia and VulcanPost, exhibits in KL and Singapore.

what didn't

The company closed in January 2016. We were brilliant at winning recognition and mediocre at winning paying venues. Every award ceremony was a week not spent closing the merchants whose counters we needed to live on.

built

Cloud printing: send a PDF from anywhere, collect it at a print shop near you. We even converted Word and PowerPoint files server-side so anything you sent came out right.

what worked

First runner-up at the IHL–MSC Startup Challenge 2016, Gold Award at ITEX 2017. The product genuinely worked, and print shops liked the extra traffic.

what didn't

Demand density never arrived. Printing is an errand people batch, not an emergency they'll pay a premium for — and the shops that were our distribution had no incentive to push a service that made them interchangeable.

built

Smart-city systems for real governments: the PJ City Bus app for Majlis Bandaraya Petaling Jaya — live bus positions and arrival ETAs for every station — and tracking analytics for solid-waste truck fleets in Petaling Jaya and Penang. I led the technical team and ran the data-science experiments.

what worked

We shipped. Two local governments ran our software in production, and my masters research in deep learning finally had a street address: real vehicles, real routes, real citizens checking when the bus arrives. The company outlived my tenure — its transit platform lives on today.

what didn't

Government sales cycles are measured in budget years, not sprints. A small team can build the product in months and then wait quarters for procurement — cashflow dies in that gap. I moved on in mid-2018; the durable business took years longer to emerge.

built

A live public Layer-1: proof-of-work consensus, 60-second blocks, a masternode-governed DAO treasury. I started by building the mining pool and ended up leading development of the chain itself — new wallet, new block explorer, the TriveAsset tokenization layer, and a consensus algorithm upgrade on a running network.

what worked

This is where I earned my protocol stripes — you learn what immutability really costs when a mistake can't be hot-fixed. Very few engineers in the region have operated an actual L1: consensus, masternodes, treasury governance. The code is still public — 29 repositories.

what didn't

A regional chain competes with global liquidity. Technical soundness couldn't overcome the gravitational pull of larger ecosystems. So in 2021 I engineered the migration of TRVC holders to BNB Chain rather than letting the network rot — a responsible sunset instead of a slow abandonment.

built

A blockchain R&D and education house — hands-on workshops, a researcher-lab program with MaGIC and the Malaysia Blockchain Association, and TRVC.App, which put durian-tree ownership certificates on-chain back in 2019 when 'real-world assets' wasn't a buzzword yet. Named Best Blockchain Technology Developer at Bloconomic Expo 2019.

what worked

Teaching compounds. The workshops and university collaborations built a network across the Malaysian blockchain ecosystem that outlived every single product I've made — and led to the Vice Presidency of the Malaysia Blockchain Association.

what didn't

Education revenue is linear: you sell hours. It funded the mission but was never going to scale like a product, and I let product work crowd out the teaching more than I should have.

built

A full game studio on Polygon, not just one game: an NFT marketplace, the SpaceDen social world inspired by Animal Crossing, Anchorland (story-driven PvE), Battlefield (PvP), and a standalone title — Space Invasion — that we later migrated into a Telegram mini-app. We even shipped Razortooth AI, an agent that talked to our community and monitored our Telegram group. Raised ~$2.5M from 21 backers at the peak of play-to-earn.

what worked

At its peak, more than 60 people worked on Playermon, coordinated through weekly meetings across time zones for genuinely global reach — a community above 140,000 followers. We kept shipping long after the hype left: marketplace V3, PvP 3.0 in 2024, the Telegram mini-app, a booth at YGG Play Summit Manila three years after launch. Through a 99% token drawdown, nobody ever accused us of running away, because we didn't.

what didn't

Play-to-earn was a subsidy dressed as a business model. When token incentives are the reason people play, the game dies the day the incentives do. We fought that gravity with real gameplay, but the model's flaw was structural. Even the ending taught me something: I exited through an acquisition, but the buyer's plans for the brand were not mine, and what Playermon became after the handover is their chapter — not mine. Diligence goes both ways: a sale is only as clean as the buyer's intentions.

built

Advisory and fractional-CTO work at the intersection of AI and blockchain — plus small, useful, ad-free tools under my own name, because craft needs an outlet that doesn't need a pitch deck.

what worked

Twelve years of pattern recognition, finally pointed at other people's ventures before the expensive mistakes happen instead of after.

what didn't

Ask me in five years. This chapter is being written — some of it live on this blog.