Gen-AI / LLM Programming

This course is given over the 4 month semester during the school year, with weekend-only and mid-week class times available. Typical weekly time commitment is 10-12 hours. 3-month accelerated cohorts available in summer months. All classes are conducted via Zoom with a live, synchronous instructor. 

After registration and tuition payment you will be given a choice of classes having different scheduled times with different weekly schedules (i.e. weekend, mid-week, morning versus afternoon, etc.)

As part of onboarding, your student student completes a brief readiness & placement review form. This review does not require prior programming experience and is not competitive. To support effective learning, students are placed into cohorts with peers of comparable prior background experience. Please be reassured, all cohorts cover the same core curriculum meeting the same academic standards. Cohorts ensure that instruction moves at an appropriate pace, while maintaining a rigorous academic environment for all groups of students, and that students are grouped roughly by age as well.

This course is equal to a bit more than a full semester of school instruction, approximately 75 hours of direct live instruction. 

Course description & overview:

• Using the LangChain framework in Python to make calls to cloud-based LLM APIs
  • Using LangChain to manage LLM prompting and prompt histories
  • Classic prompt strategies covered and use of prompt repositories
• Using Retrieval Augmented Generation (RAG) in order to search and utilize private data not present in LLM model trained knowledge 
  
  • Employing vector databases as part of RAG
  • Covering RAG best practices and design patterns to improve RAG effectiveness 
  • Querying unstructured documents via RAG such as PDFs, etc.
  • Interfacing LLMs with structured data(bases) including SQL and NoSQL or key-value stores, including schema definition using Pydantic
•  LLM Tool Calling and Agentic Workflows
  • Covering the ability of LLM's to call other software functions, programs, and tools formally, augmenting their capability with vetted tools  
  • Introduction to agents and how these have evolved into cyclic graphical methods
  • Introduction to LangGraph as a formal way of building agentic control flows
•  Using Model Context Protocol to enable standardized two-way connections between important data sources and AI-powered tools