Module 1: Introduction to AI in Programming (8 Lecture Hours)
Overview of software engineering evolution from manual coding and IDE-based development to AI-
assisted programming, fundamentals of large language models with transformers, GPT models, and
code-specific LLMs like Codex, working principles of coding copilots covering context windows,
embeddings, and training data, capabilities and limitations of LLMs in programming covering
boilerplate generation and refactoring versus challenges in novel logic and large-scale reasoning, ethical
and legal considerations covering copyright, licensing (GPL, MIT), and data leakage.

Module 2: Prompt Engineering Fundamentals (12 Lecture Hours)
Principles of effective prompts covering specificity, context, and iteration, zero-shot and few-shot
prompting for code generation, handling ambiguity through role-playing prompts such as instructing the
model to act as a senior programmer, common pitfalls including hallucinations and inaccurate outputs,
debugging prompts and iterative refinement, generating simple functions like sorting algorithms with AI,
context injection using retrieval-augmented generation (RAG) for local codebases, structured outputs for
JSON or XML responses, prompt chaining to break large requirements into sequences, negative
prompting to restrict deprecated libraries, refinement in the absence of test cases, meta-prompting for
one LLM to generate prompts for another, prompting for code optimization, reasoning-oriented
prompting with chain-of-thought and tree-of-thought techniques.

Module 3: AI for Code Generation and Autocompletion (4 Lecture Hours)
Autocompletion workflows in IDEs such as VS Code with Copilot, generation of boilerplate code
covering classes, APIs, and data structures, version control workflows with AI including automated
commit messages and branch management.

Module 4: Debugging and Refactoring with AI (6 Lecture Hours)
AI-assisted debugging to explain errors and suggest fixes, reducing cyclomatic complexity and
improving readability, refactoring loops and removing duplicated code, combining static analysis tools
like Pylint with LLM insights, handling edge cases and robustness testing through AI prompts.

Module 5: Testing and Documentation (6 Lecture Hours)
AI-generated unit tests with coverage and assertions, integration testing with dependency mocking, code
summarization, code search using embeddings, automated documentation covering docstrings,
READMEs, and API specifications.

Module 6: Building Software with LLM APIs (6 Lecture Hours) -- Optional --
Programmatic access to LLMs via APIs, prompt templates and pipelines, function calling and external
tool integration, retrieval-augmented generation for large repositories, embeddings and vector search for
semantic code retrieval, repository-level coding assistants, agentic coding workflows with multiple AI
agents collaborating to generate, test, and refine code.

Module 7: Ethical and Societal Implications (4 Lecture Hours)
Bias detection and mitigation in AI-generated code, intellectual property and ownership of AI-assisted
code, job displacement and evolving programming skill requirements, responsible AI usage guided by
ACM and IEEE standards.

Module 8: Limitations, Evaluation, and Best Practices (4 Lecture Hours)
Risks of "vibe coding" and unmaintainable AI output, AI failure scenarios with fallback strategies,
evaluation of AI outputs using code quality metrics such as cyclomatic complexity, hybrid workflows
combining AI assistance with traditional programming, future trends in multimodal AI and multi-agent
coding systems where agents generate, test, and audit software artifacts.