AI Research Engineer · Oulu, Finland
I design, build and evaluate machine learning systems: deep learning and machine vision, NLP and retrieval-augmented LLMs, and the data and evaluation work that takes them from prototype to delivery. MSc in Computer Science and Engineering (AI) from the University of Oulu, 2026, on top of four years in industry across development, consulting and technical project management.
82.7% · answer quality · thesis RAG38.5M · observations · arXiv dataset4 yrs · industry before the MSc
What I do
Selected work 4 of 16
A RAG assistant that turns corporate learning data into business insights. Hybrid retrieval (column mapping, intent detection, dense search) with two LLM judges validated against human ratings.
82.7% answer quality · +13.6 pts over baseline · MRR 0.96 · P@1 0.93
First publicly available dataset linking Finnish railway operations with synchronized weather, 2018-2024: about 38.5 million observations across 5,915 km of rail and 209 weather stations.
38.5M observations · XGBoost baseline · 2.73 min MAE
A multilingual (English, Finnish, Swedish) platform for the University of Oulu Botanical Garden: plant adoption, QR-linked plant pages, an interactive map, and an AI chat guide that answers from the garden's own data and shows its sources. Built and deployed the React front end and the chat guide, with Team Meraki.
EN · FI · SV · WCAG 2.2 AA target
A deep learning model for detecting diabetic retinopathy from retinal images, comparing neural network architectures and optimization techniques: computer vision applied to healthcare diagnostics.
Grad-CAM · attention mechanisms · ensembles
Playground toys, not work
Drop a marker on a loss surface and watch SGD, momentum and Adam find their way down.
LiveDraw a digit; a small network guesses it, running entirely in your browser.
LiveDraw a circle in one stroke; a least-squares fit scores how round it was.
LiveA thousand particles trace a hidden noise field. Your cursor stirs it.
LiveExperience