← Back to Journal

How I built my first AI project

·1 min read

Building your first AI project can be daunting. There's a steep learning curve with terms like NLP, deep learning, and transformer models. But once you break it down into manageable steps, it becomes an exciting journey of discovery.

The Problem

I wanted to build a tool that could read a bunch of resumes and automatically extract key skills and experiences to match them against a job description.

The Stack

  • Python: The language of choice for AI and data science.
  • spaCy: For natural language processing and entity extraction.
  • FastAPI: To wrap the logic into a clean, fast API.

Lessons Learned

  1. Data cleaning is 80% of the work. If your input data is messy, your model will be confused.
  2. Start small. I began by just extracting emails and phone numbers before moving on to complex skill matching.
  3. Don't build everything from scratch. Using pre-trained models from Hugging Face saved me weeks of training time.

It was an incredible learning experience that solidified my interest in artificial intelligence.