Designed a fpML-based Common Data Model serving 10+ trading channels at Westpac; built a P2P insurance platform on blockchain at Tonghui; shipped open-banking integration from zero to GA at Encompass.
generating a tailored summary…Encompass's first open-banking integration: real-time bank transaction sync into the GL with automated reconciliation rules, built from zero to GA.
A crowd-share motor insurance platform — community risk pooling backed by major Chinese underwriters — built by a 6-person founding team, funded after launch, and still operating.
The bank's first in-house CRM and an intranet portal that became the primary platform for data sharing across the Sydney office — plus Electronic Payment Services for corporate clients.
Co-founded a funded InsurTech startup in Shanghai with a team of 6, building a crowd-share motor insurance platform in collaboration with some of China's largest insurance underwriters. Exited in early 2024 after the COVID-19 pandemic made sustained Sydney–Shanghai travel impossible for three years, ultimately preventing the hands-on co-founder involvement the business needed.
Click to expand achievements
Senior consulting role delivering complex financial systems integration work for major Australian financial institutions, primarily around the Calypso trading platform.
Click to expand achievements
Interface specialist on the FXR project implementing Calypso for WIB's commodities business across front, middle, and back office.
Click to expand achievements
Seven-year tenure delivering mission-critical financial systems — from AML compliance reporting to FX straight-through processing — culminating in a project lead role owning delivery across multiple business units.
Click to expand achievements
Designed and built a crowd-share motor insurance platform — a community pooling model backed by major Chinese underwriters.
Built a crowd-share motor insurance platform where groups of drivers collectively pool premiums and share claim costs, with the risk ultimately underwritten by some of China's largest insurance companies. The business model sat between a traditional insurer and a mutual fund: members benefited from shared risk and lower premiums when the group drove safely, while the underwriter partnerships provided regulatory compliance and claims capacity. Designed the entire technical architecture and led a team of 6 to deliver it.
Conceptualised and built LBIC — a location-based on-demand travel insurance system.
Designed and implemented LBIC (Location Based Instance Cover), delivering on-demand insurance coverage to travellers based on their geographical location and current activity. The system integrated with GPS data to trigger and terminate micro-policies in real time — a genuinely novel product for the Chinese tourism insurance market. Identified the market gap, designed the product, and built it.
Delivered NAB interest derivative messaging templates for Calypso as part of Project Utah.
Collaborated with Calypso Australia to construct messaging and confirmation templates for interest derivative products (Project Utah for NAB). Also contributed to the GIC Calypso Upgrade, restructuring exchange points and interfaces for products including CIS, ELS, Variance Swap, FX, Fixed Income, and Commodity Options.
Initiated and led the Efficient Model Validation service, which progressed to pre-sale with a major AU investment bank.
Leveraged investment banking operational expertise to design a streamlined model validation service for large-scale system implementations. The initiative progressed to the pre-sale stage with a major Australian investment bank, demonstrating commercial potential beyond the immediate project scope.
Designed and implemented a Common Data Model based on fpML, standardising data across Calypso's multi-channel trading platform.
Created a Common Data Model (CDM) at middleware level to standardise data flowing in from diverse trading platforms (Baxter/Bloomberg/FlexTrade/Currenex, FXConnect, Autobahn, and Online FX). The model was based on the industry-standard fpML, maximising reusability and adaptability to new products. Delivered Credient (Sungard/Adaptive) realtime JMS interface, Oracle GL Engine, Curve update engine, and channel interfaces for FX products spanning spot, forward, swap, options, interest rate swaps, commodity derivatives, and precious metals.
fpML-based CDM serving 10+ trading channels
Built the FX Revaluation, P&L Monitoring, and Matching module used daily by Risk Control and Treasury.
In collaboration with Risk Control, Treasury Desk, and Back Office Settlements, diagnosed and streamlined FX Revaluation, FX P&L Monitoring, and FX/MM Matching processes. Built a new module for real-time data collection and calculation, enabling the business to square mismatched positions and fix process errors before settlement. The module became a core operations tool.
Built the bank's first in-house CRM (Data Pooling System) from concept to full operation.
Established the very first in-house CRM, the Data Pooling System (DPS), from initial concept through full operation. Also pioneered an Intranet Portal with Excel/PDF generation, search engine, and reporting capabilities that became the bank's primary platform for data and report sharing, improving Business Continuity by automating previously manual processes.
When we kicked off Bank Feeds, Plaid was the obvious choice — everyone knows Plaid. But I still ran a structured evaluation of Kyriba and a potential direct-connection path with Wells Fargo before recommending Plaid for GA. The lesson was that 'obvious' technology choices still need to be interrogated: Kyriba had better enterprise support SLAs; the Wells Fargo direct path would have eliminated third-party risk. Plaid won on developer experience and coverage breadth for the specific bank mix our customers use — but I wouldn't have known that with confidence without doing the evaluation. A platform recommendation that only considered one platform isn't a recommendation.
In the early 2000s at Bank of Tokyo-Mitsubishi UFJ, there was no in-house CRM. Client exposure data lived in spreadsheets and individual relationships. I proposed and built the Data Pooling System from concept through full operation — without a PM, without a design document, just a clear understanding of what the Corporate Credit Department actually needed to do their jobs. It became part of core operations. Years later that project is the clearest example I have of what it means to understand a business problem deeply enough to build the right thing without being told what to build.
At Tonghui we built a crowd-share motor insurance platform — groups of drivers pooling premiums, sharing claim costs, backed by some of China's largest insurance underwriters. Six people, no external capital at the start, a heavily regulated industry that had every incentive to ignore a startup. The underwriter partnerships were the hardest part: every architecture decision that made the platform more auditable and more obviously compliant was also a business development argument. We used blockchain for the group pooling agreements not because it was fashionable but because it made the arrangement verifiable independent of us — and that's what convinced the first underwriter to engage. The company raised external funding and continues to operate. I exited in early 2024 — COVID-19 made Sydney–Shanghai travel impossible for three years, and you can't be a hands-on co-founder of a Shanghai company from Sydney indefinitely. The exit was forced by geography, not by the business. The lesson that stuck: if you can't articulate the business case for an architectural choice, you probably shouldn't make it.
At Westpac, the FXR project had to interface with a dozen trading channels — Bloomberg, FlexTrade, Currenex, FXConnect, Autobahn, and more — each with its own data format and product representation. The solution was a Common Data Model at the middleware layer, built on fpML (the industry-standard financial products markup language). The model's design principle was that every product — spot, forward, swap, option, interest rate swap, commodity derivative — should map into a single canonical form that Calypso could consume. That one architectural decision is what made it feasible to add new trading channels later without touching the core integration logic.