Skip to content
POST/ai-integration

AI integrations

I wire language models into real applications and processes - not to impress on a demo, but to solve a concrete problem: speed up a process, support a decision, generate a useful suggestion. The first question I ask is always: what happens when the model answers wrong or slow.

User inputModel callLLMResponse validationFallback / retrywhen the model failsResult in the app

The layer around the model, not just the API call itself.

This is for you if:

  • you want to add an AI-powered feature to an existing application or process
  • you have an idea for an LLM-based product but need someone to build the technical side
  • your current AI prototype works in a demo, but you don't know how it holds up in production

What you get:

  • An integration with a language model (OpenAI, Google Vertex AI) tailored to your case, not a generic wrapper
  • A layer around the model: response validation, sensible defaults, handling errors and slow responses
  • Iterative prompt refinement based on real cases, not just ones invented at a desk
  • The AI feature wired into your backend or mobile app
example from my work

Unordered - an intelligent gift idea generator

I built an AI feature generating personalized gift suggestions in a Flutter app (Android/iOS/Web), running offline-first with data sync once the connection returns - backend on Firebase, infrastructure on Google Cloud Platform and Cloudflare Workers.

see details
stack
  • OpenAI API
  • Google Vertex AI
  • Python
  • Firebase
  • n8n