LLM-Based Web Agent Project - Prompt Engineering & Model Evaluation
Designed adaptive prompting strategies for large language models (LLMs) to optimize Planning, Acting, Memory, and Reflection capabilities. Conducted model evaluation using data collected through simulations and manual assessments. Delivered detailed reports highlighting the effectiveness of the prompting framework and its impact on task performance improvements. • Automated browser interaction and data extraction using Selenium for real-time content analysis. • Developed and tested comprehensive LLM workflows for automated task assessment. • Enhanced inference efficiency through prompt engineering and environment simulation. • Provided actionable feedback based on evaluation outcomes for future model iterations.