AI, inside the software
you already have.
I'm Ergini, an AI engineer. I wire OpenAI, Claude or open models into your existing product: streaming chat, structured output, tool-calling, retrieval over your documents, with evals and cost logging from day one. Your stack stays where it is.
A free 30-minute call, on Google Meet
Prefer email? ergin@ergini.com. I reply within hours.
What I build
A feature inside your product
Summaries, drafting, classification, search that understands the question. Behind a clean interface, so you can swap the model later without touching the rest of the app.
Retrieval over your documents
Answers grounded in your own data, with citations, hybrid search and reranking. Hosted in the EU when your data rules say so.
Tool-calling and agentic features
The model calls your APIs as typed tools, with a human checkpoint on anything that costs money or cannot be undone.
Evals, cost and observability
Per-request cost logs, latency tracking and an eval harness, so a model change is tested before it ships and the bill never surprises you.
After the call
Scope
A written plan and a price within three days.
Build
Working software every Friday.
Launch
Deployed, documented and yours.
A single AI feature usually lands between $3.5K - $15K and ships in 1 to 3 weeks; retrieval over your documents, with evaluation runs $15K - $35K. You get a written plan and a price within three days of the call.
Questions
- Do I have to rebuild anything?
- No. Integration means meeting your stack where it is. Most clients keep their API and database; I add an AI layer behind a clean interface, so the model can change without the rest of the app changing.
- Which models do you work with?
- OpenAI, Anthropic Claude, Google Gemini, and open models through Ollama, vLLM or Together when data residency or unit economics ask for it.
- What about GDPR and the EU AI Act?
- EU data residency where you need it, data minimization, a deletion path that reaches the vector store, and the logging and human oversight the EU AI Act asks for.
- How long does it take?
- A single AI feature ships in one to three weeks. Retrieval over your documents with evaluation takes three to six. You see a working demo every Friday.
- Who does the work?
- I do. No account manager, no handoff to a junior. The person on your first call writes your code, and you own all of it.
- Where are you based?
- Pristina, Kosovo, on Central European Time. I work with clients across Europe and North America, on their hours.