.NET backends and APIs
New services or rescuing slow ones: ASP.NET Core, microservices, event‑driven design on Kafka, SQL Server and PostgreSQL, deployed on Kubernetes or OpenShift.
// .NET backends · AI integrations · baked in Sarajevo
Layered like a burek. Built to hold under load.
Burekode is the independent studio of Ensar Skopljak, a senior .NET engineer with seven years in banking and payments, now shipping production LLM and voice‑AI integrations. You work directly with the person who writes the code.
var tray = await Oven.BakeAsync(
layers: 40,
filling: Filling.Meat);
// crisp outside, solid inside
await tray.ShipToProductionAsync();7 yearsbuilding .NET systems in production
300 → 100 msaverage API response on a bank payment platform past 250,000 users
9 and 11engineers in the teams I have led
Livevoice‑AI and RAG assistants running for real clients
01 / Services
Four things, done properly.
No agency layers and no hand‑offs to juniors. Each engagement is scoped around one outcome and built by me, end to end.
AI integration
Assistants that answer from your own documents and act through your systems: retrieval (RAG), tool calling into your APIs, OpenAI or Azure OpenAI, written in .NET so your team can own it.
Voice agents
Phone assistants on Retell AI that book, qualify and hand over to people, wired into your CRM and backend rather than left as a demo.
Payments and fintech
Payment gateway and POS terminal integrations, transaction flows that stay fast under load, and code that survives a bank's security review.
02 / Selected work
Systems people actually depend on.
A payment system on Kafka, three times faster
Led a team of 11 engineers and designed the payment system on Kafka. Average API response fell from about 300 ms to 100 ms while the platform grew past 250,000 users, and a project from the team was rated the best cloud application across Raiffeisen Bank International.
An assistant that answers from the client's own sources
A conversational assistant on OpenAI models with retrieval over the client's documentation, so answers are grounded in their material, and a custom tool‑calling layer that lets it act in their systems. Backend in .NET, front end in Angular, owned end to end.
A voice agent that takes real calls
Built the complete Retell AI integration in .NET for a voice assistant in production, while leading a team of 9 across AI and payments engagements, including payment gateway integrations for POS terminals.
03 / How it works
From first call to handover.
- 01
30‑minute call
You describe the problem. I tell you honestly whether I am the right person for it.
- 02
Written proposal
Scope, milestones and price in writing, as a fixed project or a monthly engagement.
- 03
Weekly increments
Working software every week in your repository, not a big reveal at the end.
- 04
Handover
Documentation and a walkthrough, so your team can run and extend what was built.
- B2B contract through a registered business
- Invoiced in EUR or USD
- CET working hours, with US East Coast overlap
- English and Bosnian/Croatian/Serbian
C# · .NET · ASP.NET Core · Kafka · SQL Server · PostgreSQL · Oracle · MongoDB · Redis · Kubernetes · OpenShift · Angular · OpenAI · Azure OpenAI · Retell AI
A slow backend, or an AI feature stuck in demo?
Tell me what you are building. You will get a straight answer within one working day.