I turn complex business processes, data and domain knowledge into working digital solutions.
From concept and domain model to architecture, AI, frontend, backend and cloud – without the overhead of a traditional development team.
Technology is the tool, not the product. Every project starts with a business problem.
Unstructured documents and information become reliable, structured knowledge – using LLMs, deterministic rules, provenance, validation and human clarification where the machine is unsure.
Calculation-intensive systems with cashflows, interest rates, scenarios, margins, risk and complex business rules – correct in substance, traceable and fast.
Processes currently spread across Excel, PDFs, email and disconnected applications become one integrated solution.
I understand complex problems, model them correctly and use modern engineering and AI to turn them into working products unusually quickly. AI is a multiplier – not the product.
Five examples – from understanding complex documents to income-and-expense accounting as SaaS.
Complex documents become reliable, structured knowledge – with provenance, rules, conflict detection and human clarification.
Contracts checked against registers, party data and internal records: conflicts of interest and contradictions become visible – and go straight into the internal review process.
Continuous learning instead of mandatory training every two years: rulebooks and new developments become short learning chapters, questions are answered, exam questions created and knowledge tested and documented – so everyone is reliably up to date.
Cashflows, interest, schedules, scenarios, margins and risk – calculation-intensive systems where domain expertise and software have to fit together.
Accounting as SaaS: quotes, contracts and invoices in a builder, compliant invoicing with e-invoices and signatures, receipts and income-and-expense accounting – in one week, mostly assembled from existing building blocks.
I help companies turn technology and AI opportunities into real business transformation.
AI transformation is not primarily a technology challenge. It requires organisations to rethink established processes, architectures, roles and ways of working. Approaches that made sense when software was expensive to build may no longer be the right ones when applications, integrations and automation can be created in days.
The biggest challenge is often not implementing new technology, but questioning existing assumptions and overcoming established patterns.
I work across strategy, organisation, architecture and execution: analysing existing models, identifying where they should be simplified or fundamentally redesigned, defining a pragmatic target state and helping teams implement the change.
My combination of business and financial domain expertise, technology architecture, engineering leadership and hands-on development allows me to connect strategic transformation with technical reality.
Rethink the model. Transform the organisation. Turn strategy into execution.
Three formats – from the first working product to technical leadership.
From business problem to working product.
Analyse the problem, establish the domain model, design the architecture and quickly build a first working version – something you can use and show, not slides.
Turn documents, rules and repetitive knowledge work into software.
Senior technical leadership without requiring a large internal technology organisation.
AI Product Architect & Engineer · mahevi
I solve problems that are difficult to understand – not just difficult to code.
My work combines business analysis, domain modelling, product architecture and software engineering – built myself, not just designed.
AI-native development now allows me to turn well-understood problems into working products in days or weeks, where traditional development often required teams and months.
Banking taught me how much compliance matters: regulation and legal requirements – from the AI Act to the GDPR – belong in the architecture from the start. Yes, they cost. But they are not a burden: in the end they protect reputation and resources.
Tools .NET / C# · Nuxt / Vue · Azure · SQL · OpenAI / LLMs · Cloud
Many processes, architectures and roles were designed for a world in which software was expensive. Introducing AI without questioning those assumptions only automates the old.
In one week, an income-and-expense accounting system with builders for quotes, contracts and invoices came together. The remarkable part is not the AI – it is that it was mostly assembly.
A language model reads values from documents remarkably well. The hard part is modelling what was read in context – and when two documents say different things, that is where it is decided whether the system can be trusted.
Small applications and experiments – evidence of breadth, curiosity and the ability to build working software quickly. Try them out.
Upload, editing and playlists — share via signed URLs.
Calculate loans and instalments, save them and share them securely.
Dynamic questionnaires and tests, distributed via signed URL.
Manage locations, images and GPS data — as a floor plan, too.
Wind at more than 60 spots worldwide, measured or forecast – the page opens on the spot nearest to you.
Loan calculator, household budget and AI assistants for open questions.
Create and send invoices and manage incoming payments — including dunning runs and export.
A simple recorder for camera and screen recording and for streaming video.
Cut recordings, add titles and subtitles, and export finished videos.
Tell me in a few sentences what it is about – process, data, goal. We will sort out the rest in a conversation.
