AI Use Policy
This page explains, plainly, how artificial intelligence tools are used in producing the content on Berojgar Engineer. I would rather tell you this openly than have you wonder about it.
1. The short version
AI helps me write. It does not decide what is true. Every technical instruction, command, configuration file and exam detail published here is something I have run, tested or checked against a primary source myself before it went live. If I have not verified it, it does not get published — or it is clearly labelled as untested.
2. What AI is used for
I use large language models (such as ChatGPT and Claude) as writing and editing assistants. Specifically:
- Structure and outlining — turning a messy set of notes from a real troubleshooting session into a readable order.
- Language and grammar — English is not my first language. AI helps me tighten sentences and fix grammar so the writing does not get in the way of the content.
- Editing and shortening — cutting repetition, rewriting paragraphs that do not read well.
- Formatting — converting notes into tables, code blocks and headings.
- Catching gaps — asking "what did I forget to explain here?" before publishing.
3. What AI is never used for
- Generating technical steps I have not performed. Commands, configuration files, error messages and fixes in the Linux, DevOps and self-hosting articles come from my own machines, my own homelab and my own work — not from a model's guess.
- Inventing facts, dates, numbers or sources. Exam dates, syllabus points, vacancy details and salary figures are checked against the official source (Lok Sewa Aayog, the hiring organisation, or the published notice) before they appear here.
- Fabricating personal experience. When an article says "I did this" or "this is what happened to me", that is my actual experience. My opinions, mistakes and career advice are mine.
- Publishing unreviewed output. No article on this site is AI output pasted straight to the blog. Everything passes through my own reading and editing.
4. How verification actually works
Before a technical post is published, it goes through this:
- Every command is run on a real system — usually a virtual machine or a box in my homelab — from a clean state.
- Configuration files are copied from the working system, not retyped from memory.
- The operating system and version I tested on are stated in the post, because instructions that work on Rocky Linux 10 may not work on CentOS 7.
- Where a step failed for me the first time, I leave the failure in the post. The mistakes are usually the most useful part.
For exam and career content, I check the current official notice or syllabus, and I say when something is my opinion rather than an official rule.
5. Why I am telling you this
Two reasons. First, a lot of what is published about Loksewa preparation and IT careers in Nepal is copied, outdated or simply made up, and readers deserve to know which kind of site they are reading. Second, this site exists to reduce information asymmetry for Nepali IT graduates — being vague about my own process would be a strange way to do that.
Using AI to write better English does not make the content less mine. Publishing something I have not verified would.
6. Things do still go wrong
Software changes, package repositories move, syllabi get revised and I make mistakes like everyone else. If you find something on this site that is wrong, outdated or does not work on your system, please tell me — it will be corrected or removed. You can contact me here or email admin@berojgarengineer.com.
7. Changes to this policy
If the way I use these tools changes, this page changes with it. It applies to all content published on https://berojgarengineer.com.