Kjetil Dyrland
Data engineer · ML engineer · Software engineerI work across data engineering, machine learning and software development, and I run Dyrland Data & AI.
Day to day I build data platforms and ML pipelines on Databricks and Azure — migrations off on-premises infrastructure, Delta Live Tables and PySpark pipelines, governance through Unity Catalog, and models taken all the way into production.
My research background is in decision-theoretic approaches to classifier evaluation, where I have co-authored two papers. Alongside that I design and ship native iOS apps: five are live on the App Store, each built end to end from the Swift client to the API and database. Most of my work is in Python, Swift, Rust or TypeScript, backed by PostgreSQL, Supabase or Cloudflare D1.
Where the production work has been.
Lerøy Seafood · Bergen Led the migration of on-premises data infrastructure to Databricks on Azure. Built production ML pipelines for spot salmon price prediction, an internal AI chatbot, and an OCR model for bill-of-lading processing. Designed pipelines with Delta Live Tables and PySpark, with governance enforced through Unity Catalog. Databricks · Azure · PySpark · Delta Live Tables · Unity Catalog
1 yr 8 moBouvet ASA · Bergen Data engineer on an Equinor project focused on big-data processing and pipeline optimisation in Databricks on Azure. Improved ETL performance and reliability across large-scale data engineering workflows. Databricks · Azure · ETL at scale
2 yr 5 moWestern Norway University of Applied Sciences · Bergen Assisted in courses on programming fundamentals, algorithms and data structures, systems development, operating systems, distributed systems and machine learning. Graded assignments and exams, and mentored students in lab sessions.
3 yr 11 moSoftware engineering and machine learning.
M.Sc. Software Engineering & Machine Learning
Aug 2020 — Jun 2022 · HVL / University of BergenGrade A. Thesis: Machine Learning in Drug Discovery, in collaboration with the Mohn Medical Imaging and Visualization Centre.
Exchange semester, Computer Science
Aug 2019 — Dec 2019 · University of California, San DiegoOne semester of computer science coursework at UC San Diego.
B.Sc. Computer Engineering
Aug 2017 — Jun 2020 · Western Norway University of Applied SciencesThesis: a cross-platform mobile application for Kronbar, on Android and iOS.
Databricks Certified Data Engineer Associate
Sep 2024 · DatabricksCertification covering the Databricks Lakehouse platform, ELT with Spark SQL and Python, incremental processing, and production pipelines.
On evaluating classifiers properly.
Two papers with Alexander S. Lundervold and P.G.L. Porta Mana, arguing that the standard way of scoring machine-learning models is not merely imperfect but avoidably wrong.
Does the evaluation stand up to evaluation? A first-principle approach to the evaluation of classifiers
Shows that popular metrics — F1-score, MCC, AUC — are mathematically never optimal: each produces an avoidable fraction of incorrect evaluations. Makes the case for grounding evaluation in decision theory with problem-specific utilities.
Don't guess what's true: choose what's optimal. A probability transducer for machine-learning classifiers
Proposes calculating probabilities from a trained classifier's output rather than from features directly — a one-time "transducer" enabling decision-theoretically optimal choices. Demonstrated on a drug-discovery problem with heavily imbalanced data.
What I reach for.
- Languages
- Python, Java, Rust, Swift, TypeScript, SQL
- Data
- Databricks, PySpark, Delta Lake, Delta Live Tables, Unity Catalog, PostgreSQL, Supabase
- Machine learning
- Decision theory, Bayesian methods, classifier evaluation, XGBoost, CNNs, MLflow
- Web
- React, Next.js, Tailwind, Vite, FastAPI
- Mobile
- SwiftUI, SwiftData, Kotlin, Xcode, TestFlight
- Cloud & DevOps
- Azure, Cloudflare, Docker, Vercel, GitHub Actions, Xcode Cloud