Bheeshm Sharma, a PhD Scholar under Prof. P. Balamurugan at IEOR, recently completed an internship as a PhD Research Sciences Intern at Microsoft, Hyderabad. During his internship, Bheeshm focused on developing an Automatic Prompt Optimization (APO) tool designed to refine system prompts for various product features.
System prompts play a critical role in shaping the interaction between users and AI-powered products, guiding the model to deliver high-quality, relevant responses. However, these prompts are often lengthy and crafted manually, which can limit efficiency and consistency.
The APO tool automates the optimization of these prompts, resulting in improved response quality while significantly reducing the number of tokens required. This not only lowers computational costs but also enhances response speed, making the overall system more efficient and scalable.
