LLMs for the Real World: Structuring Text with Declarative NLP

In our exploration of machine learning pipelines for extracting structured data from unstructured text, we'll be harnessing the power of Marvin, a lightweight AI engineering framework designed for building trustworthy, scalable, and user-friendly natural language interfaces. Here's a breakdown of the session:

1. Introduction to Large Language Models (LLMs) and their capability to transform unstructured text into structured, typesafe data.
2. The pervasive challenges engineers, analysts, and data scientists face in creating structured datasets from raw text.
3. A shift from the conventional, specialized methods: LLMs' unparalleled ability not only to extract but also to comply with distinct data models.
4. Delving deeper into Marvin's AI Models, which are rooted in Pydantic. These models merge the profound reasoning capabilities of AI with the stringent type boundaries established by Pydantic, offering developers a novel approach to NLP pipelines.
5. Real-world applications of LLMs, such as:

• Structuring electronic health records
• Crafting custom entity extraction pipelines
• Generating synthetic data for test-driven development
• Automating schema normalization processes for data warehousing.

📅 Agenda:

6:30 - 7:00 PM - Welcome

7:00 - 7:15 PM - Introductions

7:15 - 8:00 PM - Marvin Talk