Software Engineering · IT · AI · Interactive Systems
Mobile, backend, web and machine-learning research — designed, built and tested as one system.
About
Most of my projects cross a boundary: a schema decision that shapes a screen, a data pipeline that decides what a model can learn.
I work across those layers so one side never breaks the other.
SoloLearn
1Gemini Enterprise Agent Ready
Android Studio — Quail releases
Android Studio User
Compose essentialsLayered backends, relational schemas, versioned migrations.
Cross-platform mobile and static-first web, fast on poor connections.
Deep learning on physiological time-series, built to reproduce.
Vector artwork and motion authored for the browser.
Technical expertise
Hover a node to see what it was built for.
What I write production code in.
Backend services and desktop apps.
ML pipelines, ECG preprocessing, CLI tooling.
Type-safe web components and build tooling.
Typed app logic and widgets in Flutter.
Interactions that degrade to a usable page.
Queries, joins and constraints on live schemas.
Semantic structure for search and screen readers.
Grid, custom properties, container queries.
Service architecture, authenticated APIs, CI.
Layered REST services with persistence and validation.
Versioned endpoints with consistent errors.
Token auth and refresh, handled in one place.
Containerised services and reproducible dev envs.
Multi-module builds and reproducible packaging.
Branching, review and automated pipelines.
Reproduce, isolate, cover the regression.
Relational modelling, migrations, state in sync.
Schema design, indexing and query tuning.
Domain modelled in the database first.
Versioned migrations beside the service.
Auth and document data for client-led apps.
Serverless Postgres, branch per environment.
Reconciling cached content with server state.
Cross-platform clients and static-first web.
Cross-platform architecture, state and offline data.
Static sites, islands only where needed.
Small islands that need client state.
Real breakpoints from 320 px up.
Keyboard paths, focus and reduced motion by default.
Platform builds, permissions, device testing.
Deep learning on physiological time-series.
Morphology learned from raw ECG waveforms.
Hybrid models for arrhythmia over time.
Resampling and segmenting 12-lead ECG.
Subject-level splits, imbalance-aware metrics.
Model definition and reproducible runs.
One parameterised path, one variable at a time.
Vector animation built for the browser.
Scroll-linked motion across this site.
Optimised vector playback on canvas.
State-machine animation that responds to input.
Keyframes authored for Lottie, not video.
Illustration structured by what will move.
Interface design and component systems.
What I write production code in.
Backend services and desktop apps.
ML pipelines, ECG preprocessing, CLI tooling.
Type-safe web components and build tooling.
Typed app logic and widgets in Flutter.
Interactions that degrade to a usable page.
Queries, joins and constraints on live schemas.
Semantic structure for search and screen readers.
Grid, custom properties, container queries.
Service architecture, authenticated APIs, CI.
Layered REST services with persistence and validation.
Versioned endpoints with consistent errors.
Token auth and refresh, handled in one place.
Containerised services and reproducible dev envs.
Multi-module builds and reproducible packaging.
Branching, review and automated pipelines.
Reproduce, isolate, cover the regression.
Relational modelling, migrations, state in sync.
Schema design, indexing and query tuning.
Domain modelled in the database first.
Versioned migrations beside the service.
Auth and document data for client-led apps.
Serverless Postgres, branch per environment.
Reconciling cached content with server state.
Cross-platform clients and static-first web.
Deep learning on physiological time-series.
Morphology learned from raw ECG waveforms.
Hybrid models for arrhythmia over time.
Resampling and segmenting 12-lead ECG.
Subject-level splits, imbalance-aware metrics.
Model definition and reproducible runs.
One parameterised path, one variable at a time.
Vector animation built for the browser.
Scroll-linked motion across this site.
Optimised vector playback on canvas.
State-machine animation that responds to input.
Keyframes authored for Lottie, not video.
Illustration structured by what will move.
Interface design and component systems.
Selected work
The problem, the build, the trade-offs.
Deep learning research on detecting atrial fibrillation in 12-lead ECG signals, and how detection behaves as the signal is resampled from 500 Hz down to 62 Hz.
A Flutter application for working through structured theory material, backed by a Spring Boot service and a versioned PostgreSQL schema.
A Python utility for inspecting, validating and protecting PDF documents in batch, built around a scriptable command-line interface.
A Java desktop application for composing, organising and scheduling social media content from a single interface.
A production company website for HaskTolk AS with a modern frontend, service-focused content, and a secure document verification workflow.
Motion & interactive animation
Vector motion authored for the browser, self-hosted.
Move your pointer or use the arrow keys to steer
LottieInteractive
A vector driving scene illustrated in Illustrator, keyframed for Lottie export and driven by a dotLottie state machine for live steering, horn and start interactions.
Motion designInteractive
The motion system used on this site, including interactive skill diagrams, a scroll-linked experience timeline and section transitions that are cleaned up after navigation.
Research
Applied machine learning on physiological signals.
Biomedical signal processing · Deep learning
How atrial fibrillation detection holds up as a 12-lead ECG is resampled to lower rates — CNN1D against a CNN-LSTM, on PTB-XL.
Experience
Mercor · Remote
Evaluating AI-generated technical responses for accuracy and reasoning quality.
OsloMet — Oslo Metropolitan University
Deep learning for atrial fibrillation in 12-lead ECG, and what lower sampling rates cost.
Self-directed projects
Complete applications — mobile clients, backend services, databases, static sites, animation.
Freelance / Language Service Providers · Oslo, Norway
Interpreting and translation across healthcare, legal and public-sector settings.
Education
Østfold University College · Halden, Norway
Applied computer science, focused on AI, machine learning and software systems.
OsloMet — Oslo Metropolitan University · Oslo, Norway
Software development, systems, databases and networked applications.
Research topicAtrial fibrillation detection from 12-lead ECG signals using deep learning
Contact
A role, a project or a research question — email is quickest.
