Handshake AI

Improved LLM reliability in software engineering by developing hundreds of adversarial prompts designed to expose model weaknesses and edge case failures. Developed rubric-based evaluation prompts for 2D graphics and visual art assessment, creating structured criteria for measuring LLM output quality in subjective creative domains.

LLM Evaluation Prompt Engineering Software Engineering 2D Graphics

What I worked on

As part of the Handshake AI Fellowship, I worked as an AI Trainer specializing in software engineering and 2D graphics. I developed hundreds of adversarial prompts designed to expose model weaknesses and edge case failures, improving LLM reliability in real-world software engineering scenarios.

I also developed rubric-based evaluation prompts for 2D graphics and visual art assessment, creating structured criteria for measuring LLM output quality in subjective creative domains.