AI · Own product
An AI agent working for real businesses
- Situation
- Freelancers and small businesses with no time to learn AI and afraid of putting their data into a generic chat.
- What I did
- I designed and built Zina: one agent per customer on its own server, Telegram as the only channel, voice transcribed locally, memory, confirmations and daily backups.
- Result
- Customers in production across several sectors and a full restore tested in minutes.
Engineering · Platform
From own servers to AWS without stopping the product
- Situation
- An advertising platform with a PHP monolith and servers that no longer kept up with growth.
- What I did
- I dockerised the environments, broke the monolith into services and, as Engineering Manager, led the migration to AWS EKS with infrastructure as code.
- Result
- Daily deployments, an event-driven architecture and teams with a clear owner for every part.
Leadership · Process
From fixed cycles to milestone flow
- Situation
- Several product teams starting on the same day, cycles padded with filler topics and stakeholders never knowing when anything would ship.
- What I did
- With product I designed the move to Continuous Alignment: shaped milestones, a backlog per cluster, managers pulling when capacity exists and asynchronous quality gates.
- Result
- A written, approved operating model in motion, with no filler topics.
AI · Hotels
AI in hotel marketing, with judgment
- Situation
- Independent hotels with healthy demand, too much dependence on intermediaries and a new question: how does AI describe them to their guests?
- What I did
- Alongside Cometa Web I prepared the direct-sales workshop for the Tourism Innovation Summit 2026 and joined the panel on AI in hotel marketing.
- Result
- A clear thesis for the sector: AI creates value when it improves one concrete decision. And a partnership that continues.