Generative AI Boot Camp for Developers

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time budget: ~20 minutes
Section: Day 1 AM (shared with M02 lab)

Learning Objectives

After completing this lab, you will be able to:

Identify real engineering use cases where GenAI can add value
Classify each use case as automation, augmentation, or risk
Make your first LLM API call using both CLI and IDE-based tools

Tool Introductions

Before starting the exercises, familiarize yourself with the GenAI tools available on your sandbox VM. You’ll use these throughout the workshop.

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time budget: ~20 minutes
Section: Day 1 AM (shared with M01 lab)

Learning Objectives

After completing this lab, you will:

Apply the four pillars (Context, Instruction, Constraints, Format) to craft effective prompts
Iterate and refine prompts to measurably improve output quality
Generate real development artifacts using prompt engineering across multiple tools

Tool Reminders

You used these tools in M01. Quick refresher:

Tool	Best For	Command / Access
OpenCode CLI	Provider-flexible, scriptable, one-shot tasks	`opencode run "prompt"`
VS Code Chat	IDE-integrated, sees your open files	Ctrl+Shift+I in VS Code
Claude Code (optional)	Deep codebase analysis, complex refactoring	`claude -p "prompt" # Requires ANTHROPIC_API_KEY set`
Codex CLI (optional)	Fast prototyping, web dev	`codex exec "prompt"`

Today’s focus: Prompt engineering patterns work the same regardless of which tool you use. Try at least two tools across the exercises to see how each handles the same prompt differently.

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time budget: 90 minutes
Section: Day 1 PM — The Signature Lab

Learning Objectives

After completing this lab, you will:

Build a complete REST API using at least two different GenAI developer tools
Compare tools across key dimensions and identify which fits your workflow
Apply prompt engineering (M02) and the AI quality checklist to AI-generated code

Tool Introductions: Your AI Coding Toolkit

This module is about putting tools into practice. Here’s a detailed look at what’s available on your sandbox VM and when to use each.

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time budget: ~20 min | Day 2 AM (shared with M05)

Learning Objectives

Convert a user story into a complete development package using AI across the full software development lifecycle — from requirements through documentation.

Concept Review: AI Across the SDLC

AI augments every phase — but doesn’t replace human judgment

SDLC Phase	What AI Can Do	What Humans Must Do
Requirements	Expand user stories, draft acceptance criteria, identify edge cases from patterns	Validate against domain knowledge, prioritize based on business value
Design	Propose architecture, generate API contracts, suggest data models	Choose between trade-offs, ensure alignment with existing systems
Implementation	Generate code, refactor, complete patterns (M03)	Review every line, ensure business logic correctness
Testing	Generate test cases, identify edge cases (M07)	Add domain-specific tests, verify test quality
Documentation	Generate API docs, READMEs, code comments	Add context only humans know (why decisions were made)
Deployment	Generate CI/CD configs, IaC templates (M08)	Verify against infrastructure, test in staging

The Golden Rule

AI drafts → Human reviews → Human owns. Document AI involvement: git commit -m "feat: add password reset (AI-assisted)"

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: ~20 min | Day 2 AM (shared with M04)

Learning Objectives

Build a simple AI-powered chatbot API. Implement a basic LLM endpoint, then extend it with prompt chaining — a foundational pattern for AI applications.

Concept Review: LLM APIs & Prompt Chaining

Every LLM API Follows the Same Pattern

Regardless of provider (OpenAI, Anthropic, Google), the core interface is identical:

Your App → [prompt + config] → LLM API → [response] → Your App

Common parameters across all providers:

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: ~25 min | Day 2 PM (shared with M07)

Learning Objectives

Build a RAG-based knowledge assistant. Ingest documents, query with grounding, and compare RAG vs. non-RAG responses.

Concept Review: RAG in Practice

The RAG Pipeline

1. INGEST: Documents → chunks → embeddings → vector DB
2. QUERY: User question → embed → search vector DB → retrieve top-K chunks
3. AUGMENT: Build prompt with retrieved chunks as context
4. GENERATE: Send augmented prompt to LLM → grounded response

Key Components

Component	What It Does	Options
Embedding Model	Converts text to vectors	Chromadb built-in (onnx, free), OpenAI embeddings
Vector Database	Stores and searches embeddings	ChromaDB (lightweight, pip install), Pinecone (cloud)
Chunking Strategy	How you split documents	Fixed-size (200-500 tokens), sentence-based, semantic
Top-K Retrieval	How many chunks to retrieve	Start with 3-5; more = broader context but higher token cost

Why RAG Instead of Fine-Tuning?

RAG	Fine-Tuning
Dynamic — docs update instantly	Static — retrain when data changes
No model training needed	Requires training data and compute
Works with any LLM	Tied to a specific model
Provides citations (retrieved chunks)	No built-in provenance
Cheaper and faster to implement	Better for consistent style/tone shifts

Common Failure Modes

Retrieval misses relevant docs → Chunks too large or too small; try different chunk size
Irrelevant chunks confuse the LLM → Lower top-K; add relevance threshold
Context window overflow → Too many chunks; reduce top-K or chunk size
Stale data → Vector DB not re-indexed; schedule regular ingestion

Prerequisites

LLM API access (for embeddings and generation)
ChromaDB: pip install chromadb
Sample documents provided in workshop/m06/docs/

Using OpenCode or VS Code Chat: Both work for this lab. OpenCode is good for executing the Python scripts. VS Code Chat is good for explaining concepts as you go.

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: ~25 min | Day 2 PM (shared with M06)

Learning Objectives

Generate and evaluate AI-produced test suites. Learn when to trust AI-generated tests and when human judgment is essential.

Concept Review: AI-Assisted Testing

What AI Excels At in Testing

Task	AI Capability	Human Role
Unit test generation	Generates happy path + common edge cases from function signatures	Add domain-specific edge cases; verify assertions
Edge case discovery	Suggests scenarios humans often miss (null, boundary, type errors)	Filter out unrealistic scenarios
Test data generation	Creates large, varied datasets	Ensure data is realistic for your domain
Mutation testing	Introduces bugs → checks if tests catch them	Interpret results; strengthen weak tests

AI Testing Limitations

Doesn’t know your business logic: AI can test that a function returns something but not that it returns the right thing for your domain.
Can generate buggy tests: AI tests can contain their own bugs — always review assertions.
Misses implicit requirements: “This should never take more than 100ms” — AI won’t know unless you tell it.

The AI Testing Workflow

1. Provide function + context → AI generates baseline tests
2. Run tests → identify failures
3. Ask AI: "What edge cases am I missing?"
4. Add your domain-specific cases → run again
5. Review all assertions for correctness
6. Commit — human-reviewed, AI-assisted

Exercise: Generate & Validate Tests

Step 1 — Generate Baseline Tests (10 min)

Use your code from M03 or M05, or the sample code from M02 (workshop/m02/sample-code.py).

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: ~20 min | Day 3 AM (shared with M09)

Learning Objectives

Build an AI-assisted CI pipeline with automated PR review — one of the highest-ROI AI integration points for development teams.

Concept Review: AI in CI/CD

Where AI Adds Value in Pipelines

Pipeline Stage	What AI Can Do	Example
Linting	Catch semantic issues linters miss	“This function always returns None in this branch”
PR Review	First-pass automated review	Style, missing error handling, security anti-patterns
Security Scan	Explain vulnerabilities + suggest fixes	“This SQL query is vulnerable to injection. Use parameterized queries instead.”
Test Analysis	Identify flaky tests, coverage gaps	“These 3 tests fail intermittently — possible race condition”
Pipeline Config	Generate CI configs for new projects	“Generate a GitHub Actions workflow for a Python FastAPI project”

The AI-Enhanced PR Review Pattern

Developer pushes PR → CI runs AI review → AI posts review comments →
Developer addresses → Human reviewer approves → Merge

What AI catches: Style violations, missing error handling, hardcoded values, security anti-patterns, obvious bugs.

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: ~20 min | Day 3 AM (shared with M08)

Learning Objectives

Identify prompt injection vulnerabilities and implement practical guardrails — the most critical security skill for AI-augmented developers.

Concept Review: AI Security Threats

The Threat Landscape

Threat	What It Is	Real-World Example
Direct Prompt Injection	User input overrides system instructions	“Ignore all previous instructions and…”
Indirect Prompt Injection	Attack hidden in data the AI processes	Malicious instructions in a webpage the AI summarizes
Data Leakage	Sensitive data sent to external APIs	Pasting customer PII into a public LLM chat
Output Manipulation	AI output used without validation	AI-generated SQL executed without review

Defense in Depth

Layer	Technique	Example
Input	Sanitize, validate, separate with delimiters	“User query: [input] — System: respond only to the query above”
Processing	Least privilege, content filtering	Restrict AI to read-only operations; no shell access
Output	Validate before acting	Never execute AI-generated commands without review
Human	Approval for sensitive operations	Require human sign-off for: code merges, financial transactions

The Minimum AI Security Policy

Every team using AI should have a policy covering:

Lab

Mon, 01 Jan 0001 00:00:00 +0000

Time: 45 min | Day 3 PM

Learning Objectives

Build a simple agent workflow — the capstone pattern of the workshop. Understand when agents are the right solution and when they add unnecessary complexity.

Concept Review: AI Agents

What Makes an Agent?

An agent is an AI system that can reason, plan, use tools, and take multi-step actions. The key difference from a chatbot: agents don’t just respond — they act.