Software Developer, Data Annotation Online
I optimized outputs of Large Language Models (LLM) by debugging and writing code for a given prompt and its output. I created and tested hundreds of prompts designed to improve model output relating to coding and math results. I improved detection of adversarial prompts and steps to harden output against such prompts. • Debugged and tested model outputs for correctness and security. • Constructed and iteratively improved prompts across multiple programming languages. • Evaluated model responses to enhance accuracy and reduce bugs. • Contributed directly to the improvement of LLM code generation capabilities.