Built, and running.

The challenge. The client's internal team was taking too long to build a third-party KYC integration. The integration was already being billed for. Worse, electronic customer records were sitting there, available — and the onboarding team had no way to use them.

What we built. A mid-system that lets the onboarding team pull KYC records through the existing integration and export them straight into the onboarding platform.

Stack. Node.js · PostgreSQL · Redis · REST APIs · n8n · Azure App Service and Container Apps

The outcome. Onboarding became frictionless. Most documents were already in the team's hands, so clients stopped being asked for them.

The challenge. Customer and business onboarding was manual while AML and KYB regulation was tightening.

What we built. A compliance portal with OCR document extraction, liveness checks, sanctions screening, registry lookups, four-eyes approvals and an immutable audit log.

Stack. React · Node.js · Python · PostgreSQL · Azure

The outcome. Onboarding dropped from four days to under 24 hours.

The challenge. Compliance was losing hours chasing supporting documents for trades and transactions. Every request arrived after the fact. They wanted to get ahead of it.

What we built. An AI automation tool that monitors incoming transactions, categorises them, and automatically triggers document requests to the relevant client's relationship manager.

Stack. Python · FastAPI · LangGraph · Azure OpenAI Service · pgvector · Azure Batch

The outcome. By the time a compliance analyst opened the file, the RM already had the documents ready.

The challenge. Analysts recovering lost hospital revenue were working from EHR exports, spreadsheets and a shared inbox.

What we built. A case management and analytics platform with EHR ingestion (HL7 FHIR, Epic and Cerner exports), a configurable coding rules engine, and client recovery dashboards. The coders' own heuristics became rules their team can change without us.

Stack. React · Node.js · Python · PostgreSQL · HL7 FHIR

The outcome. Analysts tripled their throughput and reached first recovery three times faster.

The challenge. Property managers juggled separate tools for leases, work orders and tenant communication, with no single view of a property.

What we built. A multi-tenant, microservices SaaS platform covering the lease lifecycle, work order dispatch, tenant self-service and portfolio analytics — beginning with a month of domain discovery alongside the property managers themselves.

Stack. React · TypeScript · Python · PostgreSQL · AWS

The outcome. Lease administration time fell by 60% and most tenant requests moved to self-service.

The challenge. Race day information for athletes from more than 100 countries was scattered across PDFs, web pages and printed guides.

What we built. An offline-capable progressive web app with a live course map, targeted alerts, and a 12-language AI assistant grounded in the official race documentation rather than the open internet.

Stack. Next.js · Node.js · OpenAI · Mapbox GL · PostgreSQL

The outcome. The assistant resolved the overwhelming majority of questions without a human.

The challenge. A unified system serving several different stakeholders, each needing their own interface.

What we built. Solution architecture and advisory on maintaining a centralised data dictionary, with every system drawing its requirements from that single source. Cross-system collaboration worked by design, scaling stayed straightforward, and data fields could be changed without breaking anything downstream.

Stack. OpenAPI 3.0 / Swagger · GraphQL · PostgreSQL · Microservices · Azure API Management and Service Bus

The outcome. Duplication across the development of each interface was substantially reduced.

The challenge. A private investment company needed to reach a higher category licence and had no clear route to it.

What we built. An advisory engagement rather than a build — a complete roadmap to the target licensing category, with a step-by-step product rollout sequenced so regulatory compliance held at every stage. Regulatory framework, enterprise architecture and cyber security.

The outcome. Instead of moving blind, the company had a detailed plan for a long-term goal.

The challenge. High cloud cost and a scattered architecture.

What we built. Optimised the full stack so only required resources run, and re-architected several modules to bring recurring cloud spend down.

Stack. Azure · Terraform · Docker · Kubernetes (AKS) · Azure Monitor · Grafana

The outcome. Lower monthly infrastructure cost and higher uptime, with near-zero service failures.