Project articles
The products, infrastructure tools, and learning experiments I build. Read what each project does, how it works, and what the evidence shows.
Products and tools
Product
Read from here: your study material, out loud
Click a paragraph. Press play. Keep learning. A tiny Chrome reader with page highlighting, speed buttons, free browser voices, and optional AI voices.
Product
Track My Spend: personal finance, emotional spending, and MCP
A bilingual expense tracker combining multi-currency records, emotional spending insights, and reviewed AI imports through MCP.
Infrastructure
ClusterView: understanding Kubernetes through a live service map
A Kubernetes dashboard that connects service health, logs, dependency evidence, and optional AI diagnosis, with read-only defaults.
Product
JobScout AI: job discovery, CV matching, and evidence-based research
Inside an account-based job-search application with scheduled discovery, CV-based evaluation, and research that preserves sources and uncertainty.
Product
Ripple: personal notes with local semantic search and daily review
How Ripple combines durable notes, local embeddings, voice transcription, and a small daily review queue without generating invented answers.
Creative computing
Fly Brain Simulator: turning a connectome into a musical instrument
A browser-based neural instrument connecting a fly connectome, a designed sensory world, Web Workers, and synthesized sound.
Website
Flyttiva: a bilingual moving-service enquiry website
A focused Next.js moving-service website with a shared Swedish and English enquiry flow, server validation, and private booking emails.
Integration prototype
WhatsApp AI Agent: a tool-using assistant with LangGraph
An integration prototype connecting WhatsApp messages to calendar, reminders, notes, and optional search through LangChain and LangGraph.
Integration prototype
Job Match: a Chrome extension for CV and job-description comparison
A browser-extension prototype that compares a CV with job-page text and presents AI-generated fit analysis alongside the listing.
Learning experiments
Learning experiment
Dolphin QLoRA: why lower loss was not enough
Two local Dolphin 8B fine-tuning runs: a failed adapter, a lower learning rate, validation-based selection, and measured classification results.
Learning experiment
CNN Lab: learning image classification with PyTorch
A readable CNN project for MNIST, Fashion-MNIST, and CIFAR-10, with separate validation and test evaluation and inspectable training records.
Learning experiment
Lunar Landing RL: learning REINFORCE with a rocket
An educational PyTorch policy-gradient project with manual flight, a rule-based demo, parallel rollouts, and recorded LunarLander experiments.
Learning experiment
RL Playground: building a reusable Gymnasium basketball environment
A custom reinforcement-learning environment library with basketball physics, wind, manual play, a ballistic controller, and reusable Gymnasium interfaces.
Learning experiment
Basketball Shot RL: continuous-action policy gradients
A PyTorch REINFORCE study that learns basketball power and angle through continuous action distributions and held-out wind evaluation.
Learning experiment
Cairn Pig Lab: agent dispatch with reproducible evidence
A bounded dice-game experiment showing how Cairn assigns a task, records tournament evidence, and declares completion through one Codex worker.