Practical AI Course

Assignment: AI Timeline Research Project

Create a document (or simple Python program) covering:

• Choose 3 AI milestones that interest you and research them
• For each milestone: What happened? Why was it important? What changed afterward?
• Write about one current AI application you find fascinating
• Reflect: What excites you about AI? What concerns you?
• Optional: Create a simple program that displays your timeline with print() statements

Expected Time: 3–4 hours

Assignment: Rock Paper Scissors Game

Create an interactive Rock Paper Scissors game where you play 3 rounds against the computer. Build a complete game with score tracking, input validation, and proper result display.

Expected Time: 3–4 hours

Assignment: TicTacToe Game with AI

Create an interactive TicTacToe game where you play against the computer. Build features for board display, game logic, win detection, and a computer AI that tries to win, blocks you, or picks randomly.

Expected Time: 4–6 hours

Assignment: Weather AI Assistant Prep

Build a program that:

• Uses a free weather API (OpenWeatherMap) to get current weather data
• Stores favorite cities in a JSON file
• Has functions to: add city, remove city, get weather for a city
• Saves weather data history to a file
• Uses dictionaries to organize weather information
• Handles errors gracefully (invalid city, network issues, missing API key)
• Uses environment variables for the API key

Expected Time: 5–6 hours

Assignment: Bank Account System

Build a bank account management system that:

• Has an Account base class with attributes like account number, balance, and owner name
• Implements core methods: deposit(), withdraw(), and get_balance()
• Creates at least two account subclasses: CheckingAccount and SavingsAccount that inherit from Account
• CheckingAccount offers unlimited free withdrawals, but has monthly fees
• SavingsAccount charges a fee for withdrawals over 3 per month, but offers interest on the balance
• Uses private attributes (with underscore prefix) for balance and implements proper encapsulation
• Overrides the withdraw() method in subclasses to demonstrate polymorphism with different withdrawal rules
• Implements a Bank class that manages multiple accounts (create, find, and list accounts)
• Saves and loads account data from a file to persist information between sessions
• Includes error handling for invalid transactions (overdrafts, negative amounts, etc.)

Expected Time: 5–6 hours

Assignment: AI Writing Assistant

Build a simple writing assistant that:

• Takes a writing prompt from the user
• Calls an LLM API to generate content
• Displays the AI response
• Tracks token usage
• Allows multiple generations in one session
• Saves generated content to a file
• Lets users choose between different types of content (story, essay, email, etc.)

Expected Time: 5–6 hours

Assignment: AI Tutor Bot

Create an educational chatbot that:

• Has a well-crafted system message defining it as a patient, Socratic tutor
• Teaches a topic of your choice
• Asks the user questions to check understanding
• Provides explanations when the user is wrong (without just giving answers)
• Keeps conversation history for context
• Adjusts difficulty based on user performance
• Uses different temperature settings for different types of responses
• Implements at least 3 different prompting techniques learned

Expected Time: 5–6 hours

Assignment: Personal AI Assistant

Build a terminal-based assistant that:

• Maintains full conversation history
• Can perform multiple tasks: set reminders, take notes, answer questions, tell jokes
• Saves conversation to file for later review
• Implements a “reset conversation” feature
• Handles token limits by summarizing old messages when needed
• Has a distinct personality through well-crafted system prompts
• Allows user to choose between different AI “modes”

Expected Time: 6 hours

Assignment: Multimodal Content Creator

Create a program that:

• Takes a topic from the user
• Uses LLM API to generate a short story, article, or social media post
• Uses image generation API to create a relevant cover image
• Optionally uses text-to-speech to create audio version
• Saves all outputs to organized folders with timestamps
• Includes error handling for each API
• Lets user choose content type and style
• Tracks costs across all API calls

Expected Time: 6–7 hours

Assignment: Customer Review Analyzer

Build a tool that:

• Reads customer reviews from a CSV file
• Uses AI to analyze sentiment for each review
• Extracts key themes, complaints, and compliments
• Generates a comprehensive summary report
• Categorizes reviews by product, topic, or issue
• Identifies urgent issues that need attention
• Processes reviews in batches to manage API costs
• Saves analyzed data back to a new CSV file

Expected Time: 6 hours

Assignment: AI Email & Calendar Manager

Create a program that:

• Takes email or message text as input
• Uses function calling to extract structured information
• Categorizes messages (work, personal, promotional, urgent, etc.)
• Can add events to a simple calendar (stored in JSON)
• Can set reminders based on extracted deadlines
• Generates draft replies using AI with context
• Saves all structured data to JSON files
• Can process multiple emails from a text file

Expected Time: 6–7 hours

Assignment: Personal Knowledge Base with Semantic Search

Build a searchable knowledge base that:

• Allows adding documents, notes, or articles
• Creates and stores embeddings for all content
• Performs semantic search based on meaning
• Returns top-N most relevant documents with similarity scores
• Uses RAG to answer questions about stored documents
• Can handle at least 50+ documents
• Shows the difference between keyword and semantic search

Expected Time: 6–7 hours

Assignment: AI Research Assistant

Build an advanced research assistant that:

• Can ingest multiple document types
• Chunks long documents intelligently
• Creates embeddings for all chunks with metadata
• Implements hybrid search combining embeddings and keywords
• Uses RAG to answer complex questions requiring multiple sources
• Cites sources in responses
• Provides confidence scores for answers

Expected Time: 7–8 hours

Assignment: AI Task Automation Agent

Build an agent that can:

• Accept high-level goals from the user
• Break down goals into sub-tasks automatically
• Execute sub-tasks using appropriate tools/functions
• Maintain state and memory across multiple steps
• Self-correct when tasks fail
• Provide progress updates to the user
• Generate a final deliverable
• Include at least 5 different tools/functions

Expected Time: 7–8 hours

Assignment: Custom AI Model Project

Choose one approach:

Option A: Fine-tune a Model — Choose a specific task, create a training dataset (min 50 examples), fine-tune using OpenAI's API, and compare performance vs base model.

Option B: Advanced Few-Shot System — Build a system with dynamic few-shot examples, organized by category, using RAG to find similar examples.

Expected Time: 7–8 hours

Assignment: Deploy an AI Application

Take one of your previous projects and:

• Add a web interface using Gradio or Streamlit
• Implement proper configuration management
• Add usage logging and analytics
• Include rate limiting to control costs
• Write a detailed README with setup instructions
• Deploy to a free platform (Hugging Face Spaces, Streamlit Cloud, etc.)
• Share the deployed link and documentation

Expected Time: 7–8 hours

Capstone Project

Build a complete AI application that:

1. Solves a real problem you care about
2. Uses at least 3 AI techniques from the course
3. Has a user interface (web, terminal, or GUI)
4. Is deployed and accessible online
5. Includes proper documentation

Suggested Project Ideas

• AI Study/Research Assistant — Document processing, semantic search, RAG Q&A
• AI Content Creation Suite — Article generation, image creation, SEO optimization
• AI Business Automation — Email classification, meeting summaries, action items
• AI Code Learning Companion — Code explanation, exercises, review & feedback
• Domain-Specific AI Agent — Specialized RAG + custom tools for any domain
• Your Own Idea! — Propose a custom project

Expected Time: 10–15 hours