We work from problem to product through a process we can discuss and verify together.
I use Codex, Claude and Cursor every day to explore solutions and develop software. I compare their proposals with the requirements, review the code and verify the result in the product.
I get things done.
I listen to the people who will use the product, map the process and clarify goals, constraints and priorities.
I choose a first journey to implement and turn it into something to try. We review the result before extending it.
I review code, content and interactions, checking the original need as well as errors and unexpected situations.
I prepare a clear handover, agree the release and verify the product in the environment where it will be used.
What I learn on one project comes with me to the next. RADA brought together sensors, radio links and reliable measurements. CosmoGato meant working through the costs, limits and safeguards of an AI product. With Club Perfumes, I shaped the offer alongside the professional behind it. Each new project benefits from that experience. What I have learned, the mistakes to avoid and a closer understanding of the people who will use it.
Before choosing where to host a product, I clarify who will use it, who will maintain it and what data it needs to handle. Costs and technical constraints are part of the discussion from the start.
I compare Google Cloud, AWS and Azure with European hosting and self-managed servers against the project’s needs. I assess provider location, data region and infrastructure control separately. I also consider how to export the data and application. Provider dependencies affect future costs and maintenance.
For an AI product I compare local models with paid APIs, considering quality, resource needs and budget. I assess the cost of a reading in CosmoGato and of a video in the marketing pipeline. I add storage, outbound traffic and maintenance to understand the cost of running the whole service.
My update plan includes dependency checks, testing in a separate environment and a way to return to the previous version. I assess each alert against component exposure and user impact. The PHP fixes linked here are a concrete example of the kind of advisory to review.
The Atlantic on Black October PHP advisory GHSA-w8hw-w9fx-j79w
I distinguish service continuity from data recovery. A replica can copy an accidental deletion too. I plan backups and restore tests and agree acceptable downtime. I also assess request limits where bursts could consume resources or budget, as with AI generation.
I incorporate SEO (Search Engine Optimization), GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) into the public presentation of my projects. I work on clear content, context, sources and structure to help people, search engines and assistants understand the work.
I organise pages around people’s questions and connect answers to examples, sources and verifiable results. How I work on SEO, GEO and AEO
I use the GDPR as a reference when deciding what data a website needs, why it needs it and how long it should keep it. On this portfolio you can decline analytics and change your choice later in the footer.
Tell me what you want to build or improve, who will use it and what constraints you have. I’ll get back to you so we can work out where to start.