Back to Blog
2 min read

Learning AI as a Traditional Engineer

Three years ago, I was a cloud/data engineer. AI was “something other people do.” Now it’s a core part of my work.

Here’s how I made the transition without a PhD.

What I Thought I Needed

  • Deep math background
  • Machine learning expertise
  • Research paper reading ability
  • Understanding of neural networks

What I Actually Needed

  • API usage skills (I already had this)
  • Understanding of capabilities and limits
  • Prompt engineering (just clear communication)
  • System design patterns

The Learning Path That Worked

Month 1-2: Get hands dirty

  • Built simple ChatGPT integrations
  • Experimented with prompts
  • Made mistakes, learned from them

Month 3-4: Understand the stack

  • Learned about embeddings
  • Built a basic RAG system
  • Understood tokens and costs

Month 5-6: Production patterns

  • Error handling
  • Cost optimization
  • Monitoring and observability

Month 7-12: Advanced topics

  • Agent systems
  • Fine-tuning
  • Evaluation frameworks

What Helped Most

Building stuff. Reading isn’t enough. You have to build.

Starting simple. Don’t try to build agents on day one. Start with basic completions.

Learning from failures. Every bad prompt taught me something.

Asking questions. The community is helpful. Don’t be afraid to ask “dumb” questions.

What I Ignored

Mathematical foundations. Didn’t need to understand backpropagation to build with LLMs.

Research papers. Too theoretical for applied work. Focused on practical tutorials instead.

Perfect understanding. Started building before I felt “ready.” Learned by doing.

Advice for Traditional Engineers

You already have 80% of the skills. API integration, system design, debugging—these transfer directly.

Start with OpenAI/Azure OpenAI APIs. Don’t jump into complex frameworks immediately.

Focus on applied learning. Build things. Break things. Fix things.

Your engineering skills matter. Good AI systems need good software engineering. That’s your advantage.

The Reality

You don’t need a PhD. You need curiosity, willingness to experiment, and engineering fundamentals.

The AI part is learnable. The engineering discipline is harder to teach.

If you’re a solid engineer, you can learn AI. Start building. Start now.

The best time to start was three years ago. The second-best time is today.

Michael John Peña

Michael John Peña

Senior Data Engineer based in Sydney. Writing about data, cloud, and technology.