All projectsPersonal project
LLM Workflow Automation
A personal automation system that uses local and cloud language models to remove repetitive tasks from my daily work.
Why
A surprising amount of a developer’s day is writing the same kinds of text: pull request descriptions, GitHub issues, formal emails to clients. I wanted that handled well, automatically, and without leaking client data to the wrong place.
What it does
A set of Python scripts and prompts that plug into my everyday tooling. Depending on the privacy needs of the content, a task goes to a local model through Ollama or to a cloud model.
- PR descriptions generated from the diff and commit history.
- Issue drafting from a rough note into a well-structured ticket.
- Email formalising for client communication.
What I learned
This has been my playground for prompt engineering and local model deployment. The biggest lesson is knowing when an LLM actually improves the result and when it just adds noise.
