Calls LLM APIs
OpenAI, Anthropic, and others. Sends prompts, handles responses, manages tokens and costs.
AI ENGINEERING
You do not need a computer science degree. You need to build real systems and prove it.
The term AI engineer covers a wide range of roles. What they share is this: the ability to take an AI model and turn it into something that works in the real world.
That means API calls, data handling, deployment, and documentation. It does not mean training models from scratch or writing research papers.
Most AI engineering work today involves connecting existing models like Claude or GPT to real business problems. That is a learnable, buildable skill.
OpenAI, Anthropic, and others. Sends prompts, handles responses, manages tokens and costs.
Connects documents to AI models so they answer questions accurately from real data.
Builds systems that can take actions, use tools, and complete multi-step tasks automatically.
Cleans, processes, and routes data into and out of AI systems reliably.
Ships working systems that real users can access. Not just notebooks.
Writes READMEs, tracks performance, and iterates based on real usage.
Enough to read files, call APIs, and handle data. 2-4 weeks focused.
Build something with Claude or OpenAI. See how prompts, responses, and tokens actually work.
Connect documents to an AI. This is the most in-demand skill in applied AI right now.
Streamlit, Vercel, or Railway. Your GitHub repo needs a live link or it does not count as proof.
A document processor and a sales research agent. Now you have a portfolio. Now you can apply.
Andrew Ng courses are good. They are not enough. You need deployed proof.
Jupyter notebooks are not portfolios. If it is not at a live URL, it does not exist to an employer.
You will not feel ready. Ship something broken. Fix it. Ship again.
Certificates say you completed something. GitHub says you can build something. One matters more.