en sv
Mohammed Hammoud

Mohammed Hammoud

Senior full-stack developer building scalable web and mobile products with a strong UX focus. TypeScript, React, SolidJS, Python, Django. Based in Stockholm.

Personal notes on software architecture, frontend systems, and backend design, focused on clear explanations and solutions that work in practice.

ai astro ci django docker github-pages llm nodejs npm ollama qwen rtx5090 testing

Re-running my RTX 5090 LLM benchmark on Ollama 0.30

3 min read

Ollama 0.30 delivered dramatically better performance on my RTX 5090, especially for Qwen 3.6 35B-A3B.

Benchmarking LLM context scaling on a local RTX 5090 setup

5 min read

Measured real-world context scaling performance of Qwen 3.6 27B and Qwen 3.6 35B-A3B on a local RTX 5090 using Ollama.

Practical Ways to Reduce Token Usage in AI Agents

3 min read

Practical, real-world techniques for reducing token usage in AI agents by controlling context, input, and output.

How I Turned Codex CLI into a Structured Engineering Assistant

3 min read

Why I built reusable Codex CLI skills to add structure, discipline, and safer workflows to AI-assisted development.

Deploying to GitHub Pages with a Custom Domain

3 min read

DNS plus GitHub Pages settings that make custom domains work, and the common causes of NotServedByPagesError.

Testing Django abstract models without cluttering your migrations

2 min read

A clean pattern for testing Django abstract model behavior: temporary concrete models, SchemaEditor, and predictable teardown.

Fixing locale.Error: unsupported locale setting in Docker and CI

1 min read

The real cause of locale.Error and the two fixes that actually matter: generate locales, or standardize on C.UTF-8.

Installing npm packages from GitHub

2 min read

How to use forks and unreleased fixes without making your dependency setup hard to maintain.

Using .env variables in package.json scripts

1 min read

Shell env vs .env files: when to use dotenv-cli, cross-env, and Node's --env-file.