INFO 7375 — Prompt Engineering for Generative AI

Full Course Outline (Book-Style Structure)

(Using the agreed template for this session)

---

Part I — Foundations of Generative AI and Prompt Design

---

Module 1 — Foundations of LLMs, Prompting, and AI Literacy

Core principles of large language models, generative AI, and structured prompting. Topics include:

• What LLMs are and how they work at a high level
• Temperature, randomness, and sampling strategies
• How generative models interpret and transform instructions
• Writing your first prompts
• Understanding AI limitations and failure modes
• Prompting as human–AI collaboration

---

Module 2 — The Art and Science of Prompt Engineering

A deep dive into the mechanics, patterns, and cognitive structures of effective prompting. Includes:

• What is a prompt—and what isn’t a prompt
• Persona pattern, audience pattern, question refinement pattern
• Cognitive verifier pattern and self-checking loops
• Few-shot prompting, contrastive prompting, chain-of-thought
• Game-play and simulation-based prompting
• Structural prompt design for clarity, safety, and reliability

---

Module 3 — Prompt Engineering Patterns, Frameworks, and Architectures

Advanced prompting systems that generalize across tasks. Covers:

• The Prompt Stack (system → instruction → examples → constraints → output format)
• ReAct prompting (Reason + Act)
• Iterative refiners and prompt feedback loops
• Pattern libraries and reusable prompt templates
• Prompt compression, abstraction, and modularization

---

Part II — Integrating LLMs with Tools, Data, and Retrieval Systems

---

Module 4 — Vector Databases, Embeddings, and Semantic Retrieval

Building retrieval-augmented generation (RAG) systems. Topics:

• Embeddings: semantic geometry in high-dimensional space
• Vector databases and similarity search
• Indexing strategies and chunking best practices
• Creating powerful knowledge retrieval pipelines
• RAG failure modes and hallucination reduction
• Semantic search applications with hands-on labs

---

Module 5 — LangChain for LLM Applications

A practical introduction to LangChain for building real-world LLM-powered systems. Includes:

• LangChain architecture and core components
• Tools, Agents, Chains, and memory systems
• Integrating LLMs with APIs, documents, and external data sources
• Designing production-grade LLM apps
• Case studies of LangChain in industry
• Troubleshooting common pitfalls

---

Part III — Model Customization, Fine-Tuning, and Alignment

---

Module 6 — Fine-Tuning and Configuring Large Language Models

Hands-on and theoretical foundations for building task-specific models. Topics include:

• Transformer architecture foundations
• Pre-training: corpora, scaling laws, and compute constraints
• Instruction fine-tuning (SFT)
• Multi-task training and domain adaptation
• Evaluation of fine-tuned models
• Dataset design for fine-tuning
• Tools: LoRA, QLoRA, PEFT, Hugging Face workflows

---

Module 7 — Reinforcement Learning, Alignment, and Human Feedback

Bringing models into alignment with human intentions. Covers:

• Reinforcement Learning from Human Feedback (RLHF)
• Reward modeling and preference optimization
• Safety, ethics, and steerability
• Value alignment in generative systems
• Techniques for reducing harmful or biased outputs
• Deploying aligned models in real-world settings

---

Part IV — Agentic AI, Architectures, and Autonomous Systems

---

Module 8 — Agentic AI Systems: Theory and Practice

Understanding autonomous systems powered by LLMs. Topics include:

• What is an AI agent?
• ReAct agents (reasoning and acting loops)
• Reflexion and self-debugging agents
• Function calling and tool orchestration
• Planning and multi-step reasoning
• Short-term and long-term memory systems
• Multi-agent collaboration
• Ethical and safety considerations in agentic AI

---

Part V — Visualization, Communication, and Creative Applications

---

Module 9 — Designing Prompts for Creativity, Media, and Interactive Systems

How prompt engineering powers creative industries. Includes:

• Prompting for writing, code, games, and interactive fiction
• Prompt-driven design workflows
• Generating multimodal content (images, audio, video)
• Prompt patterns for structure, tone, and aesthetics
• Best practices for reproducible creative pipelines

---

Module 10 — Communicating AI Outputs in Academic & Industry Settings

Students learn:

• How to communicate model behavior
• How to explain outputs to non-technical audiences
• Demonstrating reasoning and uncertainty
• Documenting prompt design decisions
• Presenting models, agents, and solutions effectively

---

Part VI — Research Projects, Evaluation, and Independent Inquiry

---

Module 11 — Evaluation, Benchmarks, and AI Quality Assurance

Building a rigorous evaluation mindset. Topics:

• Model evaluation metrics
• Prompt evaluation and A/B testing
• Benchmarks for agentic AI
• Error analysis and model debugging
• Designing experiments for prompt research

---

Module 12 — Final Research Project (Capstone)

Students conduct an original research or engineering project in:

• Prompt engineering
• Agentic systems
• LLM-powered applications
• Reinforcement learning & alignment
• RAG systems and vector databases
• Evaluation frameworks
• Creative AI systems

Deliverables include:

• Proposal
• Mid-semester progress report
• Final demonstration
• Written paper and GitHub submission

---

Course Objectives

By the end of this course, students will:

Prompt Engineering & Applied LLMs

• Master prompt patterns across domains
• Build robust prompt systems for creative and analytical tasks
• Understand how LLMs interpret and transform instructions

Technical Mastery of Retrieval & Integration

• Construct scalable RAG pipelines
• Use vector databases, embeddings, and LangChain

Model Customization & Alignment

• Fine-tune LLMs and evaluate their performance
• Apply RLHF principles in practical settings

Agentic AI Development

• Build multi-step reasoning agents
• Integrate tools, memory, APIs, and planning systems

Research, Creativity & Communication

• Produce high-quality written and computational work
• Present AI research with clarity and rigor

---

Course Materials

Primary Texts

• Prompt Engineering for Generative AI — Nik Bear Brown
• How to Speak Bot: Prompt Patterns — Nik Bear Brown

Additional Materials

• Research papers on LLMs, prompting, fine-tuning, RLHF
• Industry reports on LLM applications
• LangChain documentation
• Vector DB & embeddings tutorials
• Reinforcement Learning literature

---

Course Highlights

• Hands-on focus: prompts, agents, fine-tuning, LangChain
• Cutting-edge developments in generative AI and agentic systems
• Emphasis on real-world applications and advanced engineering
• A portfolio-ready final project demonstrating mastery

---

Instructors & Approvals

• Strong Python background recommended
• Independent research expected
• No instructor approval required