8. Prompt Engineering for Technical Applications (Code, Math, Data)

By the end of this chapter, learners will be able to:

• Use prompt engineering to generate, debug, and explain code
• Solve mathematical and logical problems through structured prompts
• Analyze data, generate summaries, and simulate SQL or spreadsheet logic
• Leverage prompt styles specific to technical tasks with high accuracy

The learning objectives outline the skills students will acquire after studying this chapter. This includes using prompts effectively to generate code, debug errors, and explain programming logic. Additionally, students will learn to apply structured prompts to solve math and logic problems, analyze data, and produce summaries. The intention is to help students become proficient in leveraging AI for technical applications.

Imagine you're a chef in a kitchen. The learning objectives are like a recipe that guides you on how to use your ingredients (in this case, prompts) to create various dishes (code, math solutions, data analysis). By the end, you would not only know how to cook one dish (just writing code) but a whole menu (various applications of prompts), enhancing your versatility in the kitchen.

Language models can assist with:

• Writing and explaining code
• Solving mathematical equations
• Debugging logic
• Simulating database queries
• Creating data summaries and visualizations Prompt engineering ensures technical outputs are accurate, structured, and complete.

This chunk discusses the capabilities of language models (AI) in technical areas. They can help write code, solve math problems, debug existing code, simulate database queries, and summarize data. Effective prompt engineering is essential to ensure the outputs generated by AI are not only correct but also organized and comprehensive, thus making them useful for technical tasks.

Think of AI as a very knowledgeable assistant that can help you with your homework in various subjects. If you need help solving a math problem, the AI can guide you step-by-step. If you're struggling to write a program, it can suggest code snippets. Just like how a skilled assistant can make your learning experience smoother, effective prompts help ensure the AI understands your technical needs correctly.

Basic Code Generation

Prompt: "Write a Python function to check if a number is a palindrome." Output:

def is_palindrome(n):
    return str(n) == str(n)[::-1]

✅ Use precise language: "function", "Python", "return Boolean"

Multistep Instruction

Prompt: "Write a Python script that reads a CSV file, filters rows where age > 30, and saves the output." Use delimiters for long code:

# script goes here

Debugging and Fixes

Prompt: "Fix this error in JavaScript code:

let x = [1, 2, 3];
console.log(x[3].toUpperCase());

Explain the error." Output: "x[3] is undefined, so calling toUpperCase() throws an error."

This section delves into how prompts can both generate and debug code. The basic code generation example shows how a simple prompt can result in a functional Python script. The multistep instruction illustrates how prompts can guide the AI through more complex tasks, while debugging demonstrates the AI's ability to identify and explain errors in code.

Imagine learning to bake. You can follow a simple recipe to make a cake (basic code generation), or you might tackle a more complex dish that requires multiple steps (multistep instruction). If a mistake happens, like forgetting an ingredient, an experienced baker would help identify what went wrong (debugging). Prompts serve similar roles in guiding AI effectively through coding tasks.

Prompt: "Explain the following C++ code line-by-line to a beginner:

for (int i = 0; i < 10; i++) { cout << i; }
```"
You can also define the audience:
- "Explain to a beginner"
- "Explain like I’m a senior engineer reviewing it"

This chunk emphasizes the importance of tailoring explanations based on the audience's understanding. A well-formulated prompt can request a detailed, beginner-friendly explanation or a more technical review suitable for an experienced engineer. This adaptability enhances learning, allowing users of different expertise levels to grasp the concept effectively.

Imagine giving a presentation to a group of school kids and then to a room of experienced professionals. You’d use simpler language and examples with the kids (beginner level) while using technical jargon and industry references for the professionals (senior engineer level). Recognizing your audience helps ensure your message is clear and effective.

Prompt styles that work well:

• Chain-of-thought prompting: "Solve step by step"
• Specify format: "Give final answer in bold" Prompt Example:

"If x + 3 = 10, what is the value of x? Show your steps." Output:

1. x + 3 = 10
2. Subtract 3 from both sides: x = 7
3. Answer: 7

In this chunk, different prompt styles for solving math problems are presented. 'Chain-of-thought prompting' encourages a step-by-step method for resolving equations, ensuring that each step is clear. Specifying the format helps present the final answer distinctly, improving clarity.

Think of solving a puzzle. If you follow each step methodically (like following a chain of thought), you are more likely to arrive at the right conclusion. Specifying that the final answer should be in bold is like highlighting the main clue in a mystery novel—it ensures the most critical information stands out for easy reference.

Language models can mimic SQL or spreadsheet logic. Prompt:

"Write a SQL query to get the top 5 highest paid employees from a table named employees with columns name, salary. Output:

SELECT name, salary
FROM employees
ORDER BY salary DESC
LIMIT 5;

This chunk highlights the ability of AI to generate SQL queries as part of data analysis. By providing clear prompts, users can request specific queries to extract data from structured datasets. This capability enhances the utility of AI for data-driven decision-making.

Imagine you’re organizing a party and need to know who your wealthiest friends are. Just as you might ask a friend to list those attending based on their income, you can ask AI to write SQL commands to fetch similar data from a database. The clearer your request, the better the results, just like giving detailed instructions for the list will ensure you get everyone you want to invite.

Excel Formulas

Prompt: "Create an Excel formula to calculate the average sales in cells B2 to B10, ignoring blank cells." Output:

=AVERAGEIF(B2:B10, "<>")

VLOOKUP Explanation

Prompt: "Explain how to use VLOOKUP to match a student ID with a name from another sheet."

This section focuses on using AI prompts to handle tasks in spreadsheets, such as creating formulas. The example illustrates how to calculate averages while ignoring empty cells, demonstrating the precision of formulas. The VLOOKUP task enhances understanding of how to analyze data across different sheets.

Think of a teacher compiling grades from different classes. They would need a spreadsheet formula like AVERAGEIF to compute the average score without letting missing scores skew the results. Similarly, they might use VLOOKUP to find and match student IDs across different lists, ensuring accuracy in record-keeping.

Prompt: "Summarize this data: Product A: 40 units sold Product B: 75 units sold Product C: 25 units sold" Output: "Product B had the highest sales, followed by A and C. Total units sold: 140."

Data summarization with AI involves distilling complex information into concise statements. The prompt example shows how to provide raw sales data and receive a clear summary that highlights the key takeaways. This practice is invaluable for businesses seeking to quickly understand performance metrics.

Imagine reading a long report on your school’s sales fundraiser. Summarizing this would help you quickly understand which products were popular. Just like a summary would highlight essential points, AI can quickly distill data into insights, making it easier for busy managers to make informed decisions.

✅ Be explicit about language "Use Python" or "Write in R" avoids ambiguity ✅ Use delimiters Keeps code readable and reduces misinterpretation ✅ Ask for explanation Enhances learning, especially for beginners ✅ Define input/output "Function should take a list and return a sorted version" ✅ Use test cases "Add 2 test cases at the end of the code"

This section outlines several best practices for crafting effective technical prompts. Being explicit about the programming language prevents confusion, while using delimiters ensures that the code is formatted correctly. Requesting explanations can bolster understanding for learners, and defining the input and output helps set clear expectations for the task. Incorporating test cases at the end of code adds an additional layer of verification.

Consider planning a journey. If you're specific about the destination (like saying 'I want to go to Paris'), using maps helps avoid confusion, and having clear directions (input/output) makes the trip smoother. Just like including stopovers or rest breaks (test cases) ensures a successful trip, these best practices guide AI in producing clearer, more reliable results.

Limitation Prompt Strategy Truncating long code Break into parts, use streaming/step-by-step prompts Output formatting issues Request Markdown or use triple backticks Mistakes in logic/math Use "Think step-by-step" or request error checks Complex queries not executing Ask for explanation with query

This section addresses common limitations encountered when using AI models and suggests strategies to mitigate these issues. For example, when code exceeds model limits, breaking it into smaller parts or prompting step-by-step can aid in successful generation. For formatting issues, requesting specific output styles ensures better results. If logical mistakes occur, guiding the model to think through the problem is encouraged, while asking for clarifications helps with complex queries.

Similar to working on a challenging puzzle, if some pieces don’t seem to fit, you might take a step back, reorganize the pieces, or consult instructions. In the same way, when AI struggles with certain queries or code, using specific strategies (like breaking the task down) can lead to clearer and more accurate outputs.

Prompt engineering helps AI handle technical tasks like coding, math, and data analysis with higher accuracy and structure. By providing clear intent, language, format, and structure, you can use AI as a code assistant, data analyst, or math tutor.

This concluding chunk summarizes the core of prompt engineering in technical applications. Properly constructed prompts empower AI to effectively contribute to tasks such as coding, solving math problems, and analyzing data. Clarity in prompts maximizes the AI's capabilities, allowing it to function efficiently across various technical endeavors.

Envision training a dog. By using clear commands and rewards (structured prompts), the dog learns specific behaviors (like fetching or rolling over) better than if you were vague. Just as clear training leads to better responses from a dog, structured prompts enable AI to perform tasks more accurately, making it a valuable ally in technical projects.

--