Data Science & AI
Building by exploring
I'm developing my skills in Data Science, Machine Learning, AI and Software Engineering by turning questions and ideas into working projects.
I like projects that start with a simple question and gradually become something more interesting — whether that means discovering unexpected patterns in a dataset, teaching an application to understand natural language, or finding a better way for people to interact with a model.
This portfolio is a collection of those experiments.
Featured Projects
Half Marathon Predictor
Can a Machine Learning model understand a runner without a traditional form?
This project started with a simple idea: instead of asking someone to fill in several fields, let them describe themselves naturally.
The interesting part is what happens next. An LLM interprets the description, while a separate Machine Learning model is responsible for the actual prediction.
Voice Recall
What if your notes could be searched by what you meant, rather than by the exact words you used?
Voice Recall turns spoken notes into a searchable knowledge base.
The project explores a different way of interacting with personal information: instead of remembering the exact sentence you wrote, you can search for the idea behind it.
Offer Generator
What does a small business application look like when the business logic becomes the centre of the design?
Offer Generator began as a practical problem: creating customer offers should be simple for the user, but the rules behind those offers should remain clear and reliable.
The project became an opportunity to explore how software structure can reflect the way a real business process works.
CVTailor
Can AI help tailor a CV without giving an LLM complete control over the final document?
CVTailor explores the boundary between generative AI and deterministic software.
AI helps prepare the content, while the application remains responsible for how the final document is constructed.
More projects, more questions
The featured projects show where I'm currently heading, but the earlier projects are an important part of the story too.
Data Analysis
Before asking a model to find patterns, you first need to learn how to find them yourself.
The Titanic and Iris projects were exercises in looking at data from different angles — asking questions, visualising relationships and learning to turn observations into conclusions.
Machine Learning
Find Friends explores how Machine Learning can group people based on similarities that are not immediately obvious.
It was one of my steps from analysing data towards building models that can discover structure within it.
AI Engineering
ChatLab explores another side of AI applications: what happens when a language model becomes part of a complete application rather than just a chatbot window?
Together with Half Marathon Predictor, Voice Recall and CVTailor, it represents my growing interest in building useful systems around AI.
Half Marathon Predictor · Voice Recall · CVTailor · ChatLab
Software Engineering
Offer Generator represents another part of the same journey: learning that a working application is more than a model or a user interface.
How the code is organised, where business rules live and how behaviour is tested can be just as important as the feature itself.
What connects these projects?
The projects may look different on the surface.
One works with runners.
Another works with voice notes.
Another generates business offers.
But they all come back to the same idea:
take something messy or difficult to interact with and build a system that makes it easier to understand or use.
That's the direction I'm continuing to explore — moving from analysing data, through Machine Learning and AI, towards building complete applications around them.