FA-0483Agentic & Generative AISoftware DevelopmentDevOps, Cloud & InfrastructureData & AnalyticsMainframe & Enterprise Systems

AI for Programming

Speed Up R&D - 1 day

Introduction

Why this course

Empower code, debug, and design with AI as your co-pilot

In the modern software development landscape, efficiency and adaptability are crucial. Artificial Intelligence is not a distant concept — it is now a deeply practical tool that accelerates research, design, and development cycles. Whether you are optimizing a .NET microservice or maintaining RPG logic on an AS/400, AI can assist in ideation, code generation, refactoring, and debugging.

This course focuses on applying AI as an engineering partner, not a replacement for expertise. Participants will learn how to communicate effectively with LLMs, embed AI into real systems, and orchestrate development workflows that merge human insight with machine intelligence. The instructor — bringing over 30 years of software and systems experience — ensures that every topic connects directly to real industry needs, not academic abstraction.

Important Disclaimers

Rapid Evolution of AI Tools and Techniques The AI ecosystem evolves continuously — APIs, SDKs, local models, and IDE extensions are updated frequently. Therefore, this outline serves as a guideline framework. The instructor may adjust content, tools, and demonstrations to align with the latest stable and practical technologies available at the time of delivery.
Dependence on Technical Setup and Environment This training relies heavily on a properly configured lab setup, including AI API access, local LLM runtimes, Visual Studio Code, .NET SDK, and TN5250 connectivity for RPG. Any restrictions, such as lack of internet access, firewall rules, or hardware limitations, may require adjustments to demonstrations or exercises. Participants or IT teams are advised to coordinate setup readiness before the training to ensure a smooth hands-on experience.

Learning outcomes

By the end of this intensive one-day workshop, participants will be able to:

Understand the role of modern LLMs in accelerating R&D within software projects.
Design effective prompts (including chain-of-thought and tree-of-thought) for coding, debugging, and system planning.
Leverage AI coding assistants inside VS Code for C#/.NET 8 development.
Integrate cloud-based and on-premise LLMs into .NET and AS/400 workflows.
Automate test creation, error analysis, and feature decomposition using AI reasoning.
Apply secure and maintainable methods for embedding AI APIs into enterprise codebases.
Modernize RPG systems by bridging them with .NET-based AI integration layers.
Assess and mitigate risks such as hallucination, data exposure, and dependency on third-party AI platforms.

Prerequisites

Strong programming foundation in C# / .NET (6/7/8)
Familiarity with Visual Studio Code and REST API concepts
Access to .NET SDK, Docker, and (optionally) GPU/CPU-based LLM runtime environments
Basic understanding of RPG programming and experience using TN5250 for AS/400 connectivity
Internet access for API integration demos (if applicable)
Willingness to experiment and iterate through prompts and AI interactions

Training outline

7 modules

011. AI Foundations for Modern Developers5 topics

How LLMs understand and generate code
Overview of top AI models and where they excel
Use cases: from code generation to architecture planning
Recognizing and mitigating AI limitations (hallucinations, bias, context overflow)
Security and intellectual property considerations in enterprise codebases

022. Prompt Engineering Masterclass9 topics

Prompt structure and hierarchy: instruction, context, examples, constraints specifically for coding
Zero-shot, few-shot, and role-based prompting
Contexting techniques for coding
Project definition techniques
Planing flows and and forks
Chain-of-thought (CoT) and tree-of-thought (ToT) reasoning methods
Using “self-critique” and verification loops in AI workflows
Practical prompt templates for:
- Bug fixing and diagnostics
- Unit test generation
- Refactoring suggestions
- Task decomposition and planning
Tools for managing prompts and experiments (PromptFlow, PromptLayer, LangSmith)

033. Using AI Coding Assistants in VS Code4 topics

Setup and configuration of Copilot, Tabnine, and other modern assistants
Prompting inside code comments effectively
Evaluating AI-generated code quality and safety
Real-world demos:
- Refactoring a C# service with Copilot
- Using AI to write test cases and documentation

044. Integrating LLM APIs into .NET Applications6 topics

Overview of LLM API structures (chat completions, embeddings, function calls)
Using Microsoft.Extensions.AI abstraction for unified model integration
Implementing OpenAI and Azure OpenAI endpoints in .NET
Handling tokens, throttling, caching, and cost optimization
Demonstration: build a small AI-enhanced API endpoint in C#
Logging and evaluating prompt–response cycles for improvement

055. Local and Private LLM Integration6 topics

Why and when to run local models (privacy, compliance, cost)
Using LLaMaSharp and OllamaSharp in C# for local inference
Leveraging Microsoft Semantic Kernel for orchestration and RAG pipelines
Running hybrid (local + cloud fallback) AI strategies
Example: embedding a local inference call in a .NET service
Performance tuning and quantization options

066. AI for Planning, Debugging, and Decomposition5 topics

Using AI to transform business requirements into development tasks
Prompting for system design and module architecture
Debugging with AI: analyzing logs, stack traces, and error messages
Automated test case and regression test generation
Self-evaluation: comparing AI suggestions vs. developer insights

077. Modernizing RPG and AS/400 with AI*5 topics

Building bridges between RPG logic and .NET layers
Creating service wrappers for RPG functions via APIs
Using AI to generate code adapters, data models, and migration scaffolds
Applying AI for documentation and modernization of legacy systems
Integrating .NET AI services with TN5250 and AS/400 backends

*Subject to lab availability and participant skillsets

A programme built around your team.

Share your training goals and requirements.