Data scientist

Hi, I'm Jovan.

I work on applied AI. At Tiqets I lead the data science team. On the side I build my own products, mostly for learning languages. Before industry I spent years studying galaxies.

Portrait of Jovan Veljanoski

Career

Where I work

The common thread is applied machine learning: taking something from an idea to a system that people actually use.

Tiqets

Data scientist, now leading the team · 2021 – today

Tiqets is a curated marketplace for various touristic attractions and activities. Most of my work there sits in two areas, and I also help teams across the company see where AI can improve how they work.

AI automation

I use large language models to automate and improve operations: onboarding, curation and maintenance of our inventory, customer support and insights, and more. I also build the internal benchmarks that keep their quality measurable.

Ranking and experimentation

Deciding what customers see first: ranking and recommendation systems, and the A/B testing and measurement that tell us whether a change really helped.

Cloud Technology Solutions

2020 – 2021

Machine learning on Google Cloud for client projects, from climate control for greenhouses to computer vision for factory parts.

XebiaLabs

2018 – 2020

Predictive models for software release pipelines: working out which steps were likely to fail or run late, and explaining why, so release managers could act early. XebiaLabs is now part of Digital.ai.

Independent projects

Things I build on my own

I like making things end to end, and these days AI is part of most of them.

MyLexilingo

mylexilingo.com

A language-learning platform that I designed and built myself. Teachers and learners bring their own vocabulary, and the app turns it into exercises, conversations and stories at the right level.

It started with a simple frustration: there wasn't enough practice material for learning Macedonian, and I was learning Dutch at the time.

See all projects →

Open source

Vaex, briefly

For several years I co-founded and contributed to Vaex, a Python library for working with tables that are bigger than a computer's memory. I helped shape its machine-learning tools and gave talks about it at Python conferences.

I stepped away from it in 2023. It's an older project now and newer tools do the job better, but it's still there on GitHub. You can find my other open-source work on my GitHub profile.

Background

Before industry: astrophysics

I started out studying how galaxies form. That work taught me statistics, programming, how to work with very large datasets, and how to work on hard problems without knowing whether an answer even exists.

Postdoc, University of Groningen

2014 – 2018

I used data from ESA's Gaia satellite to find stars in the Milky Way that were born elsewhere. As part of the group of Amina Helmi.

PhD, University of Edinburgh

2010 – 2014

I studied how star clusters move around the Andromeda galaxy, to learn how it grew by absorbing smaller galaxies. Under the guidance of Annette Ferguson.