Software Engineer
I design and ship production-grade software — full-stack web apps, distributed systems, developer tooling, and clean, well-tested backends. From database schema to deployed pipeline. Zero to shipped, at velocity.
Work
Freelance · Remote — Norway (housing media)
Building an autonomous pipeline that generates consistent, on-brand imagery and video of houses and apartments — the kind of listing media published on housing websites across Norway — from raw property inputs to publish-ready assets.
Owning the system end to end: automated media generation, a reliability layer that keeps output visually consistent across a listing, and the delivery workflow that ships assets to production housing platforms.
Designing for scale and repeatability — a reproducible generation-to-delivery pipeline with observability, so new listings can be processed hands-off without a human in the loop.
Technical Skills
Languages & Frameworks
Backend & Infrastructure
AI & Machine Learning
Practices & Tooling
Work
Production-grade software — distributed systems, full-stack apps, developer tooling, and a few things built purely for the craft.
A fully decentralized AI fine-tuning framework — libp2p nodes discover each other, exchange raw PyTorch gradients across the mesh, and collectively fine-tune models without any central server. 8 phases complete. 110 tests (unit + integration + chaos). v1.0.0 on Docker Hub (linux/amd64 + arm64).
A security gateway that sits between your app and OpenAI / Anthropic / Ollama — stripping PII, blocking prompt injections, scanning responses for leaks, enforcing rate limits, and emitting a full audit trail. Drop-in FastAPI proxy, Docker-ready, Prometheus metrics.
A premium, interactive web app for exploring the periodic table and the atomic structure of all 118 elements — protons, neutrons, electrons, and shells. Features a 3D atomic viewer (React Three Fiber), guided Learn mode, and a side-by-side element Compare mode.
A lightweight project-management platform (think Jira / Linear / Trello) built as a full-stack monorepo — shared TypeScript types and validation across a Next.js frontend and an Express API, backed by PostgreSQL + Prisma, with Dockerized local infra and GitHub Actions CI/CD.
A terminal-based Pong with a self-teaching AI opponent that blends value-based (Q-learning) and policy-gradient reinforcement learning, plus opponent pattern modeling. Zero dependencies — pure Python standard library. A just-for-fun experiment that learns from every game you play.
About
I'm a software engineer who designs and builds production-grade systems — from full-stack web apps and typed REST APIs to distributed backends, developer tooling, and the infrastructure that ships them.
My work spans TypeScript & Python, React / Next.js frontends, and database-backed services — with a healthy dose of applied ML when the problem calls for it. I care about the full stack: architecture, reliability, and user-facing impact.
When I'm not shipping product, I'm deep in system design, tightening test coverage, or building things purely for the craft of it.
Environment
Hardware and tooling that power day-to-day software development and shipping.
Machine
Apple M4 Pro
14-core CPU · 16-core GPU · 24 GB Unified Memory
Runtime
Node.js · Python
TypeScript (strict) · pnpm · Bun
Data & Infra
PostgreSQL · Prisma
Redis · SQLite · Docker Compose
Daily Tools