AI Trainer - Programming, DataAnnotation

Through my 3rd, 4th and 5th years in university I’ve worked remotely as an AI Trainer on the DataAnnotation platform. The workload ranged from 5 hour weeks during busy exam periods to 40 hour weeks in my summers, though was usually around 10 hours a week.

The role mostly involves reviewing output from cutting-edge LLMs on challenging programming, logic or mathematics prompts and rating them for accuracy and functionality for the user. Some of the prompts had been given and others I had to create, giving me a chance to see where AI can be a huge support to programming or mathematical research - and also plenty of the flaws that still exist. I remember being blown away by how well one model could produce code for a translator app without much support and then amazed how poor it was at making a very basic game immediately afterward. The role also helped me learn new skills, like the fundamentals of C# and Java aswell as giving me knowledge about AI tools I’d never have come across otherwise.

Learning about AI in this very applied, in-depth way has massively informed my capcity to use LLMs in my own work in a way that doesn’t compromise accuracy or the human touch that is fundamental to Data Science and communication while allowing me to be as productive as possible.

Some key skills I improved upon were:

• Understanding the strengths and weaknesses of cutting edge LLMs
• Python
• R
• Data Visualisation
• Analytical Skills
• Technological Literacy
• Time Management