Kyle Klenk | Math to Power Industry

Coding with AI - part 3

Thu, 28 May 2026 09:00:00 -0800

Description

This session builds on the material introduced in Session 1. Here we introduce a workflow for developing applications with agentic-AI to ensure that applications source code remains understandable and transferable across workspaces. These parts will demonstrate the importance of model selection for specific programming tasks, and demonstrate some larger problems where larger models such as Claude’s Opus or Gemini 3.1 pro may be preferable to smaller lightweight models. The last part of this session will end with an introduction to locally hosted AI models and how to integrate them into programming tools, followed by some discussion on the current state of AI as it relates to intellectual property, data sovereignty, and the existential question about programmer relevancy in the job market.

Participants are again encouraged to follow along, and some time will be dedicated to allow participants to split off into breakout rooms or continue watching instructor-led examples.

Part 4

Introduce a workflow to go from an idea to an implemented project
Demonstrate how to perform code reviews
Work on larger problems where model selection has more importance for completing tasks

Part 5

Offer participants the opportunity to split into breakout rooms to implement solutions to instructor provided problems, or explore their own projects with AI-based programming tools.
For participants that wish to follow along with an instructor, they can participate in a more active development project where the instructor attempts to use their ideas
If participants prefer, there will also be the option for the instructor to demonstrate how to take an unfamiliar application (SUMMA-SUNDAILS from hydrology) with an open question of “How can we parallelize this application?” to learn how to leverage agents to explore an existing codebase

Part 6

Introduce running local AI models with Ollama
Demonstrate how to interface with Ollama using python
Demonstrate how to connect Ollama to an IDE
Discuss current topics in AI-based programming with a focus on intellectual property, data sovereignty, and the existential question about programmer relevancy in the job market

Coding with AI - part 2

Wed, 27 May 2026 13:00:00 -0800

Description

This session provides an introduction to available AI-based tools. The first block covers why developers should care about utilizing AI in their projects, followed by a brief overview of Large Language Models (LLMs) and their role in modern development workflows. The subsequent blocks follow a well-defined progression: starting with prompting web-based applications like ChatGPT, transitioning into an Integrated Development Environment (IDE) to leverage autocomplete and chat features, and concluding with autonomous AI agents.

Participants are welcome to follow along with the live demonstrations using their own coding tools. Time is reserved to walk through the setup of freely accessible tools to encourage participants to follow along and explore during the session on Thursday morning.

1. Part 1

Introductions and outline of what we will learn in these two 3-hour sessions Learning Objectives
- Navigate the Ecosystem: Select the appropriate AI tool (Browser, IDE, or CLI Agent) based on the context required for the task.
- Control the Chaos: Engineer deterministic workflows using markdown plans to constrain agents and prevent hallucination loops.
- Protect Your Data: Deploy local, offline models to utilize AI assistance without violating academic or proprietary data compliance.
Large Language Model Background Information
Demonstration using web-based LLM interface (Gemini, Claude, or ChatGPT)

2. Part 2

Introduction to Interactive Development Environments (IDE) with AI capabilities
Specific walkthrough of VSCode
Demonstration of Autocomplete and Chat features within the IDE

3. Part 3

Introduce background information on agents
Demonstration of how to use AI agents
- Implement, optimize, and benchmark a program centered around the number of ways a_n can be written as the sum of distinct positive perfect squares.
- Implement a graph building application with a graphical user interface to demonstrate agentic-AI capabilities.

Coding with AI - part 1

Wed, 27 May 2026 09:00:00 -0800

Description

1. Part 1

Introductions and outline of what we will learn in these two 3-hour sessions Learning Objectives
- Navigate the Ecosystem: Select the appropriate AI tool (Browser, IDE, or CLI Agent) based on the context required for the task.
- Control the Chaos: Engineer deterministic workflows using markdown plans to constrain agents and prevent hallucination loops.
- Protect Your Data: Deploy local, offline models to utilize AI assistance without violating academic or proprietary data compliance.
Large Language Model Background Information
Demonstration using web-based LLM interface (Gemini, Claude, or ChatGPT)

2. Part 2

Introduction to Interactive Development Environments (IDE) with AI capabilities
Specific walkthrough of VSCode
Demonstration of Autocomplete and Chat features within the IDE

3. Part 3

Introduce background information on agents
Demonstration of how to use AI agents
- Implement, optimize, and benchmark a program centered around the number of ways a_n can be written as the sum of distinct positive perfect squares.
- Implement a graph building application with a graphical user interface to demonstrate agentic-AI capabilities.