Browse
Topics
Browse the notebook by subject.
- Linux 19 The GNU/Linux system: packages, drivers, permissions and day-to-day administration.
- Networking 14 SSH, tunnels, sockets, ports, TLS and the plumbing that connects machines.
- Python 14 Scripting, automation and the local AI ecosystem.
- Web 13 Browser platforms, build tools and the modern frontend toolchain.
- AI 12 Local models, transformers, speech and text generation — experimentation in the open.
- Docker 12 Containers, images, Compose and Swarm — building, running and debugging containerised services.
- Shell 7 Bash, tmux, dotfiles and the command-line craft that turns a workstation into a workshop.
- Security 6 Reconnaissance, hardening and the defensive habits that keep systems trustworthy.
- JavaScript 5 The language of the web, its runtimes and its sharp edges.
- Nix 5 Reproducible development environments and declarative tooling.
- Node.js 5 Server-side JavaScript, scripts and the small automation utilities around a project.
- Tooling 4 Editors, language servers, shells and the small utilities that sharpen a workflow.
- Transformers 4 Hugging Face models and pipelines for translation, generation and classification.
- Vue 4 Composition API, components and pragmatic frontend architecture.
- CSS 3 Layout, typography and the craft of styling interfaces.
- Git 3 Version-control workflows, history archaeology and the commands worth remembering.
- GPU 3 Graphics cards, drivers and the hardware acceleration behind local AI.
- TTS 3 Text-to-speech: turning written text into audio, on the CPU or the GPU.
- TypeScript 3 Typed JavaScript: generics, patterns and the small abstractions that scale.
- Hardware 2 Laptops, adapters, batteries and getting the most out of physical machines.
- Kubernetes 2 Containers at scale, kubectl and the control plane.
- Traefik 2 A modern reverse proxy and its Docker and Compose integration.
- Audio Processing 1 Raw PCM, buffering and windowing: turning a stream of chunks into something a model can classify.
- Benchmarking 1 Reproducible harnesses that turn assumptions about speed and accuracy into numbers.
- CI/CD 1 Pipelines, automation and shipping software reliably.
- Community 1 Forums, asking good questions and learning from other people's answers.
- CUDA 1 NVIDIA's parallel compute platform for running models and workloads on the GPU.
- Devbox 1 Reproducible, per-project development environments built from a simple devbox.json.
- DSH 1 DeepSeek Harness, agent presets and building an AI-assisted engineering workflow.
- Engineering 1 System design, measurement and the trade-offs behind software that has to keep running.
- FastAPI 1 Building typed Python APIs and the documentation that ships with them.
- Immich 1 Self-hosted photo and video management, and how to make it look the way you want.
- Java 1 The JVM, the JDK and running Java applets or applications.
- Meta 1 Notes about the notebook itself — its history, scope and housekeeping.
- Ollama 1 Running large language models locally and wiring them into everyday developer tooling.
- Open Source 1 Maintaining public software: releases, contributor reports, coordinated disclosure and the community around a project.
- Oremi 1 The personal-assistant ecosystem: how its audio, wake-word and text components fit together.
- Performance 1 Profiling, hot paths and the discipline of measuring before reaching for an optimisation.
- PocketSphinx 1 The small offline speech recogniser at the heart of the Oremi wake-word engine.
- PouchDB 1 An offline-first JavaScript database that syncs, and how to make it behave in a bundler.
- TensorFlow Lite 1 In-process model inference: interpreters, thread counts and the XNNPACK worker trap.
- Wake Word 1 Keyword spotting — making a short phrase like “oremi” fire reliably, in more than one language.
- WebSocket 1 Full-duplex streaming between clients and a long-running server — and the concurrency it forces.
- Whisper 1 OpenAI's speech-recognition models for transcribing audio locally.
- XMPP 1 The open messaging protocol behind Jabber and Ejabberd.
- YAMNet 1 Google's audio-event classifier, and what it costs when it runs through TensorFlow Lite.