%20scale(.46)%20translate(-32%20-32)'%3e%3cpath%20d='M4%2038%2024%2032'%20stroke='%23ffffff'%20stroke-width='3'%20stroke-linecap='round'/%3e%3cpath%20d='M32%2012%2052%2048H12z'%20fill='%23ffffff'%20fill-opacity='.14'%20stroke='%23ffffff'%20stroke-width='3'%20stroke-linejoin='round'/%3e%3cg%20stroke-width='3'%20stroke-linecap='round'%3e%3cpath%20d='M41%2030%2060%2021'%20stroke='%23ef4444'/%3e%3cpath%20d='M41%2031%2060%2028'%20stroke='%23f59e0b'/%3e%3cpath%20d='M41%2032%2060%2035'%20stroke='%2322c55e'/%3e%3cpath%20d='M41%2033%2060%2042'%20stroke='%233b82f6'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
Olle Oltorp
M.Sc. Machine Learning · Industrial Engineering and Management, KTH
Education
M.Sc. Industrial Engineering and Management (Machine Learning)
2025–Jun 2027 (expected)KTH Royal Institute of Technology · Stockholm
Grade 4.70 / 5.00Relevant courses: Machine Learning, Artificial Neural Networks and Deep Architectures, Deep Learning in Data Science, Scalable Machine Learning and Deep Learning, Search Engines and Information Retrieval Systems, Artificial Intelligence, Computer Security
B.Sc. Industrial Engineering and Management (Computer Engineering)
2022–2025KTH · Stockholm
Relevant courses: Software Engineering, Language Engineering with Introduction to Machine Learning, Introduction to Computing Systems Engineering, Programming Techniques, Applied Computer Science, Database Technology, Internet Programming
- Built the collaborative filtering model, which ranked newsletters best of the four systems compared (precision@5 0.568), by factorising a joint article–newsletter matrix so that article read time could predict newsletter engagement.
- Scoped with the client what their data could support and whether article personalisation was worth extending to newsletters; presented the results to around 30 of their engineers and managers.
Experience
Teaching Assistant
2026–2027KTH, School of Electrical Engineering and Computer Science · Stockholm
Research Assistant
Apr–Aug 2026KTH, Dept. of Industrial Economics and Management · Stockholm
AI adoption
WASP-HS- Synthesised the recurring barriers to AI adoption across 33 empirical studies into a thematic review, by building the coding framework with the project leads and coding every study against it.
Sustainable food
Digital Futures- Produced KTH’s first measure of uptake of its sustainability standard by analysing 2025 catering data (16,773 order lines, SEK 11.3M).
- Sole-authored the 38-page report, showing that a standard only holds where knowledge of it and authority over the choice sit with the same person; based on ten semi-structured interviews I designed and conducted.
Admissions Data Assistant
Feb–Mar 2026KTH, Dept. of Industrial Economics and Management · Stockholm
- Validated 44 confidential applications to KTH’s international master’s programmes: recomputed grades, verified university rankings.
Property Maintenance Worker
2022–2025SKB · Stockholm
- Re-hired for four summers, one autumn and one winter.
Projects
Product Search Study
2026 · Open source- Raised search quality on a 43,000-product catalogue from nDCG@10 0.710 (BM25) to 0.804 at 37 ms p99, by testing 11 embedding models, 120 hybrid setups, 15 rerankers and 1,530 LLM expansion setups, significance-tested on held-out queries.
Minecraft Mob Object Detection
2026 · Open source- Built a 49,861-image dataset of all 87 mobs (177,312 boxes and masks) labelled exactly by the game through a mod I developed, and fine-tuned RF-DETR on it to 0.602 mAP50-95 on test images from unseen parts of the world; published both on Hugging Face.
Marketplace Pricing Extension
Adds rarity grades and cross-market prices to marketplace listings.
2026– · LiveMarketplace Discovery Tool
Grades tens of thousands of live listings against community standards.
2026– · LiveGrading Data API
Public REST API powering both tools above, serving source-cited grading data.
2026– · LiveWorkout Tracker
Phone-first app for planning workouts, logging sessions and tracking progress.
2026– · LiveSole developer on all four, full stack to ops.
Skills
- Programming
- Python, JavaScript, SQL, Java, Docker, Git
- ML & data
- PyTorch, Hugging Face (Transformers, Datasets, Hub), Scikit-learn, Pandas, Polars, Matplotlib
- Search & NLP
- BM25, Dense and hybrid retrieval, Rank fusion, Rerankers, LLM query and product expansion, Embeddings, Elasticsearch
- Vision
- Object detection (RF-DETR), Synthetic data generation
- Recommenders
- Collaborative filtering, Matrix factorisation, Content-based and hybrid systems, Implicit feedback
- Evaluation
- nDCG, Recall, mAP, Dev/test discipline, Latency budgets
- AI workflows
- Multi-agent development, Model-role delegation, Task decomposition, Cost optimisation
- Research
- Semi-structured interviewing, Qualitative coding and thematic analysis
- Spoken
- Swedish (native), English (fluent)