My degree is in Computer Engineering Technology, not Computer Science. I never sat through a formal software engineering curriculum, never learned proper design patterns in a classroom, and never worked on a team with senior engineers reviewing my pull requests. And yet a live, production B2B e-commerce store exists today, built mostly with AI-assisted development, because I learned how to work with these tools properly instead of just typing questions into a chat window.
Where this actually started
My capstone project at Northern Technical University was a CNN-based traffic-sign recognition system, built in Python. That gave me a real, hands-on foundation in how machine learning models actually work, which mattered later, but it did not teach me production software engineering. Between 2021 and 2023, I built five-plus Telegram bots and small n8n automations freelance, using AI tools to accelerate development, prompting, debugging, and iterating on working code without going through a traditional software engineering workflow. That is where the muscle actually got built.
DAD LINK: the proof point
The clearest example of what AI-assisted development can actually deliver is DAD LINK, a live B2B e-commerce site for network cabling equipment that I designed and shipped on Odoo, the open source e-commerce and ERP platform, with Claude Code doing most of the heavy lifting. That included the product catalog structure, category pages, on-page SEO, blog content on cabling standards, and the storefront layout itself. It is a real production store today, listing authorized distributors across Baghdad and Erbil, not a demo or a portfolio mockup.
Why "just prompting" is not what actually happened
People who have not built anything real with AI tools tend to assume it is just typing a clever request and copying the output. Shipping something that stays online, handles real traffic, and does not quietly break in three weeks takes more than that. I think about it as four layers, and I go deeper into this on the blog in the four-layers post:
- Prompt engineering: phrasing the instruction clearly, with the right constraints.
- Context engineering: making sure the model sees the right prior decisions, files, and product details at the right time, not just the last message.
- Harness engineering: the guardrails and validation around what the model is allowed to do and how its output gets checked before it ships.
- Loop engineering: letting an agent run for a long stretch, checking its own work, and correcting course, rather than babysitting every single step.
Where a CS degree would have helped, and where it would not have
I will not pretend a formal software engineering background is worthless; it would have given me cleaner instincts around architecture and testing discipline earlier. But it would not have changed the core skill that actually mattered here: knowing how to give an AI model the right information, the right guardrails, and the right validation loop to produce something trustworthy. That skill is not gated behind a specific degree. It is built by doing real work, under real deadlines, and paying attention to what breaks.
What this means if you are hiring or building
If you are trying to decide whether AI-assisted development can actually deliver a production result for your business, not a proof-of-concept, DAD LINK is the answer, not a hypothetical. I would rather talk through your specific project than make abstract claims about what AI tools "can do" in general. You can read more about how I got here, or just reach out directly if you have something that needs building.