Build, debug, and optimize AI agents using DSPy and MLflow.

Complete introduction to the simplest, most efficient, and yet most powerful way I’ve found to create AI agents, AI workflows, and AI programs in Python. Instead of manual prompting, we use automatic prompt optimization with DSPy and its concept of signatures. Timestamps / Outline: 00:00 How to Call LLMs from Python, the Simple Way 0:21 Declare Your First AI Program (in 1 LOC) 2:24 Setting Up Your Large Language Model Backend 6:10 Program 2: Processing Images 9:14 Deeper Dive into Signatures 14:01 Program 3: Processing Entities from Paragraphs 19:19 Fetching text from wikipedia with Attachments 20:39 Setting Up a DataFrame 22:22 Apply Gemini Flash lite to each paragraph 23:02 Creating a Synthetic Gold Set 24:35 Quick Baseline Evaluation 25:11 Creating DSPy Examples 25:55 Evaluation Metric 26:10 Prompt Optimization with DSPy 29:10 Final Evaluation Follow Max: Twitter: [https://x.com/MaximeRivest](https://x.com/MaximeRivest) GitHub: [https://github.com/MaximeRivest](https://github.com/MaximeRivest) Links to Relevant Repositories: Attachments: [https://github.com/MaximeRivest/attachments](https://github.com/MaximeRivest/attachments) DSPy: [https://github.com/stanfordnlp/dspy](https://github.com/stanfordnlp/dspy) FunnyDSPy: [https://github.com/MaximeRivest/funnydspy](https://github.com/MaximeRivest/funnydspy) Docs: [https://dspy.ai/](https://dspy.ai/) [https://maximerivest.github.io/attachments/](https://maximerivest.github.io/attachments/) If you’re new to my channel, my name is Maxime Rivest. I’m an Applied AI Engineer and Data Engineer. I like to educate people on the best tools in Data Analytics and AI Engineering. Max

A library for graph deep learning research. Contribute to divelab/DIG development by creating an account on GitHub.