How to improve your CV for AI and data roles

Generic CV advice ('use action verbs,' 'keep it to one page') applies here too, but AI and data roles have a specific failure mode: candidates list tools and techniques without ever describing the actual problem or outcome. A reviewer can't tell the difference between someone who genuinely built something and someone who followed a tutorial.

Replace tool lists with outcomes

'Built a recommendation system using PyTorch and collaborative filtering' tells a reviewer what libraries you used. 'Built a recommendation system that increased click-through on the product page by identifying patterns the previous rule-based system missed' tells them what problem you solved. Lead with the second, and let the first follow as supporting detail.

Quantify what you can, honestly

Numbers are useful when they're real: a dataset size, a latency improvement, an accuracy gain, a percentage of manual work removed. If you genuinely don't have a number, describe the scope instead ('processed the full customer support backlog') rather than inventing a metric that won't survive a follow-up question in an interview.

Structure your projects section like a short case study

• One line on the problem you were solving and why it mattered.
• One or two lines on your specific approach and what made it non-trivial.
• One line on the outcome or what you learned if the project didn't ship.

Link to something real

A GitHub profile with a couple of finished, documented projects is worth more than a bullet point claiming familiarity with a framework. If your best work is private or under NDA, a clean personal project, even a small one, gives a reviewer something to actually look at.

Keep the formatting boring

Most companies still run CVs through an applicant tracking system before a human sees them. Stick to standard section headers, avoid text inside images or complex tables, and use a common file format. This isn't the place to be creative. Save that for the portfolio.