Optimizing the dev workflow with LLMs

Rajika Imal

I'm extremely bullish on being more productive with LLMs such as ChatGPT and GitHub Copilot. I've started producing more output in a meaningful manner with these tools.


  • Helps to find answers directly without navigating links. However as a developer you can't blindly trust LLMs. This is where you come in as a the human interpreter of the results.
  • Unit tests.
  • Regex (HUGE!)
  • Optimizing functions

GitHub Copilot

  • Boilerplate code
  • Straighforward function logic suggestions (I guess this is part of boilerplate code?)
  • Drafting complex functions

Pretty excited about Copilot X and AWS CodeWhisperer.

Prompt Engineering