Selected work

What we have shipped, and for whom

Our engagements sit inside licensed platforms and operational environments, so what we can publish is limited by the agreements we sign. What follows is what those agreements allow.

Book a whiteboard session

Core and payments

Shaype

An Australian financial product innovation partner, operating under its own AFSL. Brands, businesses and banks use its ecosystem to build, test and launch payments products without becoming a licensed institution themselves.

How we work in this domain

93 ms

against a 200 ms requirement

No downtime

the move from a SaaS core to Shaype Core in 2021

Three months

in production in parallel

The engagement

Two periods of work on Shaype's licensed platform. Since 2023, as Sofitect, we have worked for Shaype's end clients, which we do not name, helping them integrate with the platform. In 2023 Mobiquity, a systems integrator since acquired by Hexaware, contracted us to build the platform's webhook integration layer for mobile phone notifications. From 2024 to 2025 we built a data analytics and prediction system alongside Shaype's own engineers. In 2026 we integrated DT One, a perks supplier for mobile top ups, gift cards and rewards, and Beforepay, a micro lending platform. From 2018 to 2023 our managing director, Marius Telespan, was Shaype's Head of Engineering. The team he led helped build Shaype Core, a skinny core built as an event sourced ledger, and the reconciliation around it, moved the platform onto Shaype Core in 2021 from the SaaS core it ran on, with no downtime, and built the authorisation path that decides a card transaction inside its 200 ms requirement.

Sofitect, since 2023: integration of Shaype's end clients with the platform
Sofitect, 2023: the Shaype webhook integration layer for mobile phone notifications, under contract to Mobiquity, a systems integrator now part of Hexaware, for a Shaype end client
Sofitect, 2024 to 2025: a data analytics and prediction system for a Shaype end client, built alongside Shaype's own engineers
Sofitect, 2026: mobile top ups, gift cards and rewards (DT One), for a Shaype end client
Sofitect, 2026: micro lending (Beforepay), for a Shaype end client
2018 to 2023, as Head of Engineering: Shaype Core, a skinny core, which the team helped build
2018 to 2023, as Head of Engineering: the move from a SaaS core to Shaype Core in 2021, customers, accounts, cards, transactions, authorisation holds and ledgers, with no downtime after three months in production in parallel
2018 to 2023, as Head of Engineering: 93 ms at the 95th percentile from a card transaction reaching the API to the authorisation decision, against a 200 ms requirement, with card preferences, limits by payment type, balance and fraud checks in one pass

Two periods, and we say which is which. Sofitect has worked on Shaype's platform since 2023, for Shaype and its end clients; we do not name those clients because they are Shaype's, not ours, and the webhook layer was built under contract to Mobiquity, the integrator then serving one of them. Shaype Core, built as the event sourced ledger, the 2021 move onto it from a SaaS core, and the 93 ms authorisation path were all the work of the team our managing director, Marius Telespan, led as Shaype's Head of Engineering from 2018 to 2023, before Sofitect existed. That is his track record, and Shaype has agreed to its appearing here. We say so because a reference call would establish it anyway, and because it is why we know the platform from the inside.

Fraud and risk

Undisclosed client

An Australian bank offering deposits and lending. Listed on APRA's register of authorised deposit taking institutions (ADI), and holder of an Australian Financial Services Licence (AFSL) and an Australian Credit Licence (ACL): broadly, a prudentially regulated bank with financial services and consumer credit permissions.

How we work in this domain

The engagement

Adding fraud and risk decisioning to a bank that is already clearing: Sardine on inbound and outbound rails, designed to go in without a maintenance window. In delivery.

ADI, AFSL and ACL, Australia
Sardine fraud and risk decisioning on inbound and outbound account to account rails, in delivery
Shadow scoring first, then a staged switchover

Two regulatory regimes, the same engineering

The licensed platforms we work on are Australian and in production: a banking as a service platform operating under its own Australian Financial Services Licence (AFSL) and a bank on APRA's register of authorised deposit taking institutions (ADI). The rails have different names; the engineering does not. An ADI is a prudentially regulated deposit taker, broadly a PRA authorised bank. An AFSL is broadly an FCA authorisation. AUSTRAC's regime for money laundering and terrorism financing asks the questions the UK Money Laundering Regulations and JMLSG guidance ask. APRA CPS 234 sits where DORA and the FCA and PRA operational resilience rules sit.

What was built for those regulators is what a UK bank, EMI or payment institution has to build for theirs: an authorisation decided inside the scheme window, a ledger reconciled against settlement files, fraud decisioning on inbound and outbound rails, reports traced to ledger entries.

Beyond finance

Ocean Infinity, Governr, iSense Solutions and FixDay. The same engineers, on problems a bank would recognise: pipelines sized for a continuous workload, an inventory of AI models, predictions with a confidence score, a disputes portal for investigators.

Marine robotics · UK

Ocean Infinity

A marine robotics company that combines software, AI, data and robotics to change how operations at sea are run, replacing crewed vessels with remotely supervised fleets.

The engagement

Two projects. Real time 3D data visualisation for ocean and subsea datasets, with pipelines sized for continuous survey workloads and interfaces built for operators rather than for slideware. Then a platform that predicted asset failures and faults aboard the vessels.

AI governance · UK

Governr

An AI risk management product that discovers and governs an organisation's AI assets: systems, models, datasets, agents, APIs and MCP servers. It exists to contain security, IP, privacy and regulatory exposure.

The engagement

The platform behind the product, for a UK client: front end, back end and infrastructure, with our team embedded alongside theirs. It is the kind of inventory of models, agents and datasets that the PRA's model risk management principles (SS1/23) now expect a bank to keep.

Research and data · Romania

iSense Solutions

A Romanian market research company producing consumer insight and data driven strategy for brands.

The engagement

We built the platform iSense uses to move from fieldwork to insight faster. Collection, cleaning and analysis happen in one place, and a machine learning engine does the heavy computation: predictions and patterns from the data, each with an accuracy and a confidence score, so a researcher knows how far to trust a result before it goes to a client.

Marketplace · Romania

FixDay

A Romanian startup whose platform helps customers find qualified tradespeople.

The engagement

We built the platform. The hard part was matching: each tradesperson would only travel a set distance, so the right job had to reach the right person, which meant geolocation and geofencing over a lot of data. We also built the operations portal for disputes, where investigators uncover the details of a job and resolve it.

Leave with the approach, the risks and what it would take

Ninety minutes with the engineers who would do the work, on one problem you name: a processor replacement, an onboarding flow, a ledger reconciliation.

Book a whiteboard session See the work