Updated June 2026

AI-Assisted Software Engineering Fundamentals

Class Duration

7 hours of live training delivered over 1-2 days to accommodate your scheduling needs.

Student Prerequisites

At least one year of professional software development experience
Familiarity with version control (Git)

Target Audience

Software engineers, engineering managers, and tech leads who want to use AI tools productively without accumulating technical debt, security vulnerabilities, or eroding code ownership. Equally useful for teams rolling out AI assistants organization-wide and for developers who have already started using AI tools but want a principled framework for when and how to rely on them. Pairs well with Prompting and Context Engineering for Software Engineers for the prompt-craft foundation.

Description

AI coding assistants accelerate development - but only when used with discipline. This course builds the professional habits and mental models that separate developers who leverage AI effectively from those who ship AI-generated bugs. We cover the full cycle: understanding what AI assistants are good at and where they confidently produce wrong output, establishing a code review discipline for generated code, maintaining authorial ownership, handling security-sensitive contexts, designing team workflows and standards, and measuring the true productivity impact. Labs use realistic codebases and realistic failure modes.

Learning Outcomes

Identify the categories of tasks where AI assistance provides reliable value vs. where it requires extra scrutiny.
Apply a structured review checklist to AI-generated code before committing it.
Recognize common AI error patterns: hallucinated APIs, incorrect logic, subtle security issues, and license concerns.
Design team-level workflows: prompt libraries, .instructions.md files, review gates, and onboarding standards.
Evaluate when to override or disable AI suggestions (security-sensitive code, compliance scope, novel domains).
Measure productivity impact honestly using appropriate leading and lagging indicators.

Training Materials

Comprehensive courseware is distributed online at the start of class. All students receive a downloadable MP4 recording of the training.

Software Requirements

Students should have a GitHub Copilot, Cursor, or Claude Code subscription active in their preferred IDE for hands-on labs.

Training Topics

What AI Assistants Are Good At

Pattern-heavy, well-documented problem domains
Boilerplate reduction and mechanical transformation
Test scaffolding and documentation drafting
When models are confidently wrong

Review Discipline for Generated Code

Code review checklist for AI-generated output
Logic correctness, edge cases, and error handling
API hallucination detection
Dependency and version verification
Maintaining authorial understanding

Security Pitfalls

Common vulnerabilities in generated code (OWASP relevance)
Injection risks in generated queries and templates
Secret handling and credential leakage
When to turn AI assistance off entirely

Team Workflows and Standards

Shared prompt libraries and instruction files
Pull request policies for AI-assisted changes
Onboarding developers to team AI standards
Balancing autonomy and consistency

Code Ownership and Quality

Preserving architectural intent with AI assistance
Avoiding over-generated complexity
Refactoring AI output to match team style
Long-term maintainability considerations

Measuring Productivity Impact

What to measure: cycle time, review iteration, defect rate
Leading indicators vs. vanity metrics
Team surveys and developer experience data
Communicating results to leadership