AI/ML Engineer · RAG & LLM systems · MSc Computer Science & Engineering (AI), University of Oulu
AI/ML engineer with four years of industry experience before an MSc in Computer Science & Engineering (AI) at the University of Oulu. I build retrieval-augmented and LLM systems and machine-learning pipelines in Python, PyTorch, and PySpark, on a foundation of shipping software and leading technical delivery.
- Co-authored the first publicly available dataset linking Finnish railway operations to synchronised weather data (~38.5M observations, 2018-2024); established an XGBoost baseline of 2.73-minute MAE for station-level delay prediction. arXiv:2601.16592
- MSc thesis: a retrieval-augmented assistant that turns corporate-learning data into business insight. Answer quality 82.7% (+13.6 points over the manual baseline); retrieval MRR 0.96, Precision@1 0.93. Open-access thesis
| Project | What it is | My role |
|---|---|---|
| recipe-rag-nlp | Retrieval-augmented search over 500K+ recipes (BM25 + Sentence-BERT + hybrid). NDCG@k 0.95, MRR 1.0. | Sole author |
| BloomOulu | Multilingual digital platform for the Oulu Botanical Garden with a source-cited RAG chat guide (GrowthHack 2026). | Built the React front end and the AI chat guide. Team project with @hsn07pk and others |
| steam-games-analysis-bdpa | PySpark analysis of 97,405 games; Random Forest explaining 69% of variance in success. | Sole author |
| diabetic-retinopathy | Modular deep-learning pipeline (Grad-CAM, attention, ensembles) on DeepDRiD. | Sole author |
| UBISS 2026: Robotics | Leader-follower multi-robot system using only ultrasonic sensing (UbiComp Summer School). | Implemented the control logic in CircuitPython |
- Integrating Meteorological and Operational Data: A Novel Approach to Understanding Railway Delays in Finland. arXiv preprint (cs.LG), CC BY 4.0, Jan 2026. arXiv:2601.16592
Based in Oulu, Finland · open to relocation