consulting & engineering · since 2010
20+ years of experience, 20+ international enterprise-grade clients, 20+ high-volume applications delivered. IFRS and HU GAAP expertise, steady fintech grounding, payment integration and process development & automation.
economic event → judgement → calculation → accounting posting → statutory reporting → business insight — every figure traceable in between.
our own calculation engine, ifrCoworker, covers recognition, measurement and derecognition. each result comes with its workings, Dr/Cr journal entries citing the paragraph, the disclosures it triggers and XBRL/ESEF tags — and carries on into every related standard.
statutory accounts under national rules, and the adjustments that turn them into IFRS figures.
EBA FINREP and MNB supervisory reports, from the datamart to the submitted template — implemented for 40+ credit institutions.
reports built on the taxonomy itself, so every figure carries its tag and is checked before filing.
posting rules and interfaces that carry every sub-ledger event into the general ledger without manual rework.
one batch run computes the financial impact of each line item for the reporting period along the related standard chain, following the item’s lifecycle; calculation sheets and working papers come from the same inputs.
the convenience of an AI agent, held to a deterministic engine — ending in a substantive result: an IFRS posting, an Excel report, or an appointment booked from the agent window. the language model chooses the step, the engine does the calculation, every step is logged — a dependable co-worker and quality reviewer.
your systems published as tools that Claude and other agents can call — per-user authentication, typed inputs, validated before they run.
assistants inside finance and operations workflows that prepare, check and flag — a person signs off where it matters.
3–4 coding agents under one analyst’s direction — one builds, others review and test. a production-ready MVP in 2–5 days.
acceptance criteria written precisely enough to serve as prompts, so an agent builds — or recognises and measures under IFRS — one verifiable block at a time.
anomaly and fraud detection, forecasting and classification on your own data — always compared with a simple baseline first.
radically simplified, slim solutions — the complexity stays in the system, not on the user’s screen.
requirements, BPMN process models and functional specifications — agreed by finance and IT before a line is coded.
customer- and staff-facing applications behind millions of customers: bank-branch front-ends in four countries, a retail loyalty and customer-service platform with points settlement, a national consumer portal for e-invoices and online payment.
full-stack web applications and REST APIs in Node.js, Java, Python and SQL — deployed on GCP or Clever Cloud and run in production.
data lakehouse on Azure/Databricks: self-service BI, faster management reports, and governed archiving with rehydration.
card payments and payment orchestration, e-invoice presentation, camt.053 bank-statement matching and automated AR posting — Revolut, SimplePay, corporate bank APIs.
document exchange moved from e-mail to signed digital workflows: e-signing, validation against the EU trusted list (eSignVerification), archiving.
ERP and core-banking implementations and upgrades: data migration, module set-up, general ledger and bank interfaces.
multi-team scrum, backlog and release management through to live support — led by a Professional Scrum Master II.
test management, UAT coordination, and automated and load testing with Playwright and JMeter.
repetitive back-office steps automated and monitored — every run reports what it processed and flags what it could not.
your existing APIs and data made available to AI agents over MCP and A2A — typed, authenticated and logged.
selected enterprise engagements — described, not named.
our own products, live — the capabilities above at work.
