[P3-Research] Explore DSPy for automated prompt optimization

[Ztrimus]

Overview

DSPy (Declarative Self-improving Python) can automatically optimize prompts against evaluation datasets.

How it works

1. Define task signature: (job_description, resume) -> tailored_resume
2. Provide 20-50 labeled examples of good tailoring
3. DSPy tries different prompt strategies and selects the best
4. Optimized prompts are saved and reused

Potential for ResumeFlow

• Optimize the section-wise prompts automatically
• A/B test different prompt strategies
• Adapt prompts per industry/role type

Prerequisites

• Evaluation metrics ([P2-Testing] Add DeepEval evaluation metrics for resume quality scoring #57)
• Labeled examples of good resume tailoring