Software & AI Engineer · Pristina, Kosovo

Altin Murtezaj

I build software, and the AI that makes it genuinely useful.

I design and ship products end to end, from the data model to the interface, and bring in LLMs where they actually earn their place: search, assistants, and automating the work people shouldn't be doing by hand. A background in data analytics means I care as much about what goes into a model as what comes out of it.

Open to software and AI engineering roles

Focus: Full-stack products, LLM applications
Stack: TypeScript · Python · .NET
AI: LLM integration, RAG, agents
Based in: Pristina, Kosovo

01About

An engineer who builds with data and AI.

I'm Altin, a software engineer who builds full products, and increasingly, the AI inside them.

I build software end to end: the data model, the backend, the interface, and everything it takes to get it running in production. More and more of that work involves AI, whether it's wiring an LLM into a product, grounding it in real data with retrieval, or building agents that take repetitive work off people's plates.

I came to engineering through data. Years of writing SQL, modelling messy datasets and building reports taught me to distrust anything I can't verify, which turns out to be exactly the instinct you want when working with models. Building is also what I do for fun. Ideas pile up faster than I can ship them, and every so often one becomes a real, working product.

Start with the problem

AI is a tool, not a feature. I work backwards from what someone is actually trying to get done, and only reach for a model when it does that job better than plain code would.

Get the data right

A model is only as good as what it's given. Clean data, a sound schema and well-grounded context matter more than which model sits on top. That lesson came straight from my years in analytics.

Ship, then measure

I'd rather put a working version in front of people and learn from it than polish a demo. Real usage, logs and evaluations tell me what to fix next.

02Skills

Engineering is the foundation. AI is the edge.

The first list is how I bring AI into products. The second is the engineering toolkit underneath it, including the data background I came from.

AI Engineering

How I make software smarter.

LLM integrationAPIs, tool use, structured output
RAGEmbeddings, retrieval, grounding
AI agentsMulti-step workflows and automation
Prompt engineeringReliable, testable prompts
EvaluationMeasuring what the model actually does
Vector searchSemantic search over real data
PythonServices, pipelines, automation
pandas / NumPyData preparation and analysis

Software Engineering

What I build products with.

Frontend

React
Next.js
Angular
TypeScript
Tailwind CSS

Backend

.NET / C#
Node.js
Express
Java / Spring Boot
REST APIs

Databases

PostgreSQL
SQL Server
MongoDB
DynamoDB

Data

SQL
Power BI
Tableau
Data modelling

Tooling

Git
Docker
AWS
Postman
Figma

03Projects

Things I've built, and why I built them.

Each of these started with a real problem someone had, usually one I found interesting enough to keep working on after hours.

Themeli

A construction marketplace for Kosovo

The problem: Buying construction material in Kosovo still runs on phone calls and word of mouth. There's no single place to compare what a supplier stocks, what it costs, or whether the business on the other end is one you can trust. Finding a plumber or a civil engineer works the same way.
What I built: A marketplace with three sides to it: a catalogue of materials and tools from verified suppliers, a searchable directory of professionals by trade and city, and a request-for-quote flow where someone posts what their project needs and suppliers come back with offers. An admin panel sits behind all of it for supplier verification, moderation and orders.
My role: Built it end to end: the data model, catalogue and search, the RFQ flow, the verification pipeline, and the admin dashboard. That last part is where both halves of what I do meet: I shipped the product, then built the reporting layer to actually understand what was happening inside it.

Next.js
Node.js
PostgreSQL

LindiDoors

Order and production management for a door factory

The problem: A door manufacturer was running orders on paper and spreadsheets. Nobody could say which order sat at which production stage without walking the floor and asking.
What I built: A CMS and CRM that digitises the whole order lifecycle: customer records, specifications, inventory and production stages, all with role-based access so the office, the workshop and management each see what belongs to them.
My role: Designed the data model and built the full stack.

React
Node.js
Express
PostgreSQL
JWT

AgFerm

Production tracking and analytics for agriculture

The problem: Producers were tracking fermentation cycles in notebooks. Nothing could be compared across batches, so the same mistakes repeated season after season.
What I built: A platform for recording every cycle, monitoring it live, and reading the results back as dashboards, so a producer can finally see which conditions actually produced the better batch.
My role: Full-stack build, including the reporting and analytics layer.

React
Node.js
PostgreSQL
REST API
Dashboards

Food Delivery

Ordering with real-time delivery tracking

The problem: Restaurant ordering falls apart at the moment the customer wants to know where their food is, and the kitchen wants to know what's next in the queue.
What I built: A full-stack ordering platform with live order state pushed to both sides, complex menu handling, and map-based tracking from the kitchen to the door.
My role: Built the frontend and the .NET backend, including the real-time layer.

React
.NET Core
SignalR
MSSQL
Google Maps API

Frontend repo Backend repo

Zabeli Architects

Portfolio site with a self-service CMS

The problem: An architecture studio needed its work presented properly, but every update meant going back to a developer.
What I built: A restrained, image-led site backed by a custom admin panel, so the studio adds and reorders projects themselves without touching code.
My role: Designed and built both the public site and the admin dashboard.

React
.NET
AWS DynamoDB
Admin panel

Frontend repo Backend repo

GlobESIM

Buying and activating eSIM data plans

The problem: Getting mobile data abroad still means hunting for a shop. eSIM solves it technically, but the buying experience is usually the hard part.
What I built: An e-commerce flow wired directly into telecom provider APIs, so a plan bought online is provisioned and usable within seconds of payment.
My role: Worked on the storefront and the provider and payment integrations.

React
Node.js
Payment gateways
Telecom APIs

Repository

04Contact

Say hello.

Whether it's a software or AI engineering role, a question about something I've built, or an idea you want a second opinion on, I read everything that comes in.