Looking to make a career change to working on artificial intelligence (AI), but intimidated by coding? The new field of prompt engineering is an excellent path to explore.
As someone reading this guide, you’re probably a tech enthusiast taking a smart step by exploring how to adapt to the new reality of technology but may not think they’re up for being a programmer. We can guide you through exactly what you ought to know. TL;DR: In addition to gaining an understanding of and practicing the art of prompt engineering, we recommend that you learn the basics of Python, but you certainly don’t need to become an AI engineering expert.
In this guide, we’ll get you up to speed on what prompt engineering is, the responsibilities of a prompt engineer, and the skills and knowledge you need to get the job.
Jump to a section:
- What is prompt engineering? Definition, roles, and job outlook
- What skills do you need to become a prompt engineer?
- 7 routes for becoming a prompt engineer in 2024 and beyond
- Frequently asked questions
- Accelerate your career in prompt engineering with CodeSignal Learn
What is prompt engineering? Definition, roles, and job outlook
Prompt engineering is the practice of figuring out what to say to AI models like ChatGPT so you can get the results you want. Seems simple, so how is that a whole job?
There are two levels to prompt engineering. One is simply being thoughtful about what you type as your AI prompt, whether you’re composing it yourself or copy-pasting a template. This is fast becoming a relevant skill in many roles—up there with, say, proficiency in MS Excel. The other, which is our main focus in this guide, is setting up systems that guide and enhance users’ input for optimal results. It’s like the difference between managing a budget and being an accountant.
Professional prompt engineering is a great fit for creative thinkers who are technology enthusiasts. Once you’ve learned enough Python, you’ll be able to apply your understanding of the world and way with words to get the most out of the AI models.
Prompt engineer role & responsibilities
Prompt engineering considers the entire context of a person’s interaction with AI.
- User interaction design is the first step. This could be as simple as thinking through how to suggest users write the prompts that you’ll manipulate, but it can also involve working with product designers and others to adjust the interface, flow, and other aspects of the user experience.
- Reformulation is the process of editing a user’s input so they get a better result from the AI model. You do this with scripts that restructure the user’s entry, add context, and give instructions on qualities like tone and length—in other words, engineering the prompt.
- Testing is the iterative process of evaluating results and adjusting inputs in an effort to improve them. Interestingly, this is also how AI models improve.
- Bias, fairness, and ethics is not prompt engineering’s responsibility alone, but the role serves as the front line of detecting and mitigating issues that may arise from large language models (LLMs) based on a very wide range of content and viewpoints.
Not all roles will include all four of these responsibilities, and there likely will be others you’ll have to do too. Given how new the title is, you can expect that what you do will change often.
What does a prompt engineer work on?
Generally, prompt engineers are hired to get the best possible outputs from AI for a particular use case.
These days, many new sites and apps, as well as functions within existing sites and apps, create rich output despite minimal user input. This input can be explicit, such as “10 slides of an itinerary for a trip to Croatia” in a slideshow app, or implicit, such as a writing assistant that automatically reviews your copy for tone and clarity.
If that one sentence prompt or any given paragraph of your writing went directly to the AI, it probably wouldn’t give a useful result. That’s where reformulating comes in, framing the input with plenty of context and instruction to give the model much more direction. In the slideshow app use case, it could end up as a multi-stage, behind-the-scenes iterative process, where the first prompt asks for an outline, then further ones guide the AI to flesh it out, while another process prompts an image-generation AI to create visuals.
What qualifications do you typically need to become a prompt engineer?
Prompt engineering is a new field with widely different responsibilities and requirements. But generally, in addition to becoming adept at the practice of prompt engineering, you’ll need to know some Python and demonstrate a fairly strong understanding of the fundamentals of how generative AI works.
Can you become a prompt engineer without a degree?
Some—but definitely not all—prompt engineering roles require or prefer a degree in computer science (CS) or a related topic. Even then, the employer might be flexible because CS majors are in high demand.
Whether you need a college degree of any sort depends on any given employer’s requirements. As with most knowledge-work roles, having a degree may be more important if you’re earlier in your career; with a demonstrated record of success, many will evaluate you on your performance rather than your formal education.
Can you become a prompt engineer without previous experience in AI?
The familiarity you gain with AI on the road to becoming qualified as a prompt engineer is a good starting point. You don’t necessarily need explicit AI-related job experience to be hired as a prompt engineer, but you should certainly be able to demonstrate both on your resume and in interviews that you have familiarity and competence with AI well beyond having used ChatGPT a few times.
What kind of companies employ prompt engineers?
Prompt engineering roles are popping up at organizations throughout the AI universe, from the very companies that develop the models to businesses ranging from healthcare to consumer finance.
Prompt engineering roles at high-flying tech companies, especially the ones developing AI models such as OpenAI and Anthropic, are more likely to require deep programming skills. If you’re just getting started with Python, you may want to look at places applying AI more narrowly, such as for a specific app or a business goal. Many agencies and consulting firms are hiring prompt engineers to develop workflows for individual clients that are looking to capitalize on the recent explosion in AI.
As with most knowledge work, there’s nothing inherent to prompt engineering that requires you to work in an office, and accordingly some of these roles are remote. Others are hybrid or fully on site, but that’s a matter of company culture and policy rather than the nature of the role.
What skills do you need to become a prompt engineer?
In brief: Get up to speed on how LLMs work, learn some Python, and understand and practice various forms of prompt engineering.
Understanding AI, ML, NLP and LLM
First, let’s define these terms you’ll see all over the place:
- Artificial intelligence (AI) is a broad term for technology that simulates how the human brain learns and processes information. Any one instance of AI is a model. AI is also used as a term for any given model, such as, “The AI will respond to your prompt.”
- Machine learning (ML) is, well, the learning part of AI. Almost all ML takes place through an iterative learning process, using a ton of math to get incrementally better at generating the best output for a given input. Most ML leverages deep learning, where neural networks with multiple interconnected layers work together to help computers learn complex patterns from data.
- Natural language processing (NLP) is how computers understand and generate human communication. It’s a core component of translation, speech recognition, and any other software that reads, listens, or talks.
- Large language models (LLMs) are the new hotness, forming the basis of generative AI (gen AI). ChatGPT, Gemini, Claude, Llama, and all these other super-powerful AI models have been trained on absolutely enormous amounts of text (like, pretty much the entire internet), and through a ton of processing, have created an exquisitely complex understanding of just about everything humans talk about—and how they talk.
At their core, LLMs are very sophisticated prediction machines. Given any input, they work by guessing what the next word ought to be—and the next, and the next, and so on, until their best guess is that they should stop talking. They don’t actually “know” anything in the ways humans do, but rather they do a (usually) pretty good job of guessing what to say and how to say it.
Two implications of how LLMs work are very important for prompt engineers to understand:
- LLMs are not deterministic. That means, given the exact same input, you’ll probably get a different output every time. This is how LLMs can create output that sounds more human than anything else computers have previously generated, but also makes it really hard to get consistent outputs.
- The models can make stuff up. This is known as hallucination. They don’t automatically fact-check themselves, so there’s no guarantee that their predictions lead to accurate results.
Communication skills
Compared to most software engineering roles, a prompt engineer needs strong communication skills for two audiences: the AI and your colleagues.
You’ll have to have a strong command of written English and be able to work with a variety of speaking styles and rhetorical structures. The phrasing and approach you use to deliver prompts to the AI will influence the output, so you ought to be able to try different ones.
Similarly, because you will be working with a range of technical and non-technical colleagues, you’ll need to clearly articulate your perspective to help them understand both the value of your role as well as your particular concerns and needs.
Scripting and programming skills
While not all prompt engineering roles list knowing how to code as a requirement, many do.
What programming language should I learn?
If you don’t already know how to code, Python is your best bet for getting started. It’s the most commonly used language for working with AI, and it’s relatively easy to learn. Plus, LLMs are very good at generating and critiquing code, so you can improve your skills with the help of the same AI that you’ll be engineering prompts for.
Other languages are valuable too, depending on the context. JavaScript is the backbone of modern web development, so it’s likely to come up when interfacing with the front end. Others like SQL and R are great for databases and data analysis. If you already know one or more other languages besides Python, that’s fantastic and worth highlighting on your resume.
Learning tip: Want to learn the basics of Python? The Introduction to Programming with Python in CodeSignal Learn takes from writing your first program in Python to mastering functions and more.
Understanding how to avoid biases in AI responses
AI can make offensive or biased statements. They’ve learned from hundreds of years of literature, plus the entire Internet, and people have said and implied all sorts of racist, sexist, and otherwise biased things. While AI platforms themselves try to filter out unsavory responses, it’s important to stay attuned for subtler issues or even, sometimes, big ones that slip by the controls.
The best way to avoid bias is to remove biased material from the corpus that AI is learning from, but that’s a herculean task at scale and one you probably won’t have influence over anyway. Instead, you’ll have to test and iterate to make sure that the prompts you’ve engineered can render acceptable outputs based on a wide range of user inputs. You might also be able to guide the AI in critiquing itself, though there is no substitute for human judgment in these matters.
Understanding the different prompting techniques for guiding language models
Prompt engineering can achieve much better outputs by guiding how AIs think. Sometimes this can be done in a single prompt, other times it’s a series of prompts. Here’s a quick rundown of some of the top techniques:
- Chain-of-thought prompting: Encourage the AI to think step by step, leading to more logical and detailed responses. By guiding the AI through a sequence of thoughts, you can achieve more accurate and comprehensive answers.
- Context setting: Provide the AI with the right amount of context to understand your queries better. These instructional prompts ensure that the AI has all the relevant information it needs to generate appropriate and insightful responses.
- Few-shot learning: Use examples within your prompts. By including a few demonstrating the type of response you’re looking for, you can significantly improve the quality of the AI’s responses, even with minimal initial data.
- Creative prompting: From generating novel ideas to crafting engaging stories, this technique helps you harness AI’s creativity by framing prompts in imaginative and open-ended ways.
7 routes for becoming a prompt engineer in 2024 and beyond
Set yourself up for success as a prompt engineer by not only learning the fundamentals of that skill, but also developing a solid understanding of how today’s AI works.
Participating in online courses and bootcamps
Learning prompt engineering and coding through online platforms is the sensible, affordable way to go. You can start anytime, work at your own pace, and take advantage of online exercises that make it as easy as possible to practice your new skills.
Prepare for a career in prompt engineering with CodeSignal Learn
The CodeSignal Learn platform is built for upskilling, which means it’s great for preparing for an AI career by learning prompt engineering. What’s more, you can use it for free.
Learning tip: Get started with Prompt Engineering for Everyone in CodeSignal, and if you like what you see, continue to learn the basics of Python.
Learning the fundamentals of NLP, including Python NLP libraries and frameworks
Understanding the core notions of natural language processing (NLP) will help you become a more successful prompt engineer. Although the inputs and outputs of an LLM are words and sentences that make sense to humans, what goes on behind the scenes is a whole bunch of math.
Fortunately, you don’t need to know linear algebra; there are many NLP libraries and frameworks that give you a whole lot of power with a few lines of code. See the Text Classification with Natural Language Processing learning path for an example of what you could soon know how to do.
Understanding the fundamentals of ChatGPT and transformer models
If terms like gen AI, transformer models, and backpropagation through time are foreign to you, then you ought to spend some time studying the fundamentals of modern AI.
We’ll give you a head start: transformers are a type of neural network that use a technique called attention to consider the entire context of a given input while generating an output. Transformers are the basis of gen AI models such as ChatGPT, and they use a technique called backpropagation through time to improve, by reflecting on how they came to a certain output and making adjustments to do better next time.
Many AI applications these days take a pre-made transformer—GPT stands for generative pre-trained transformer—and use a technique called fine-tuning to focus and improve its performance on the basis of additional data.
Learning tip: Have some coding experience and want to dig deeper into how AI works? Learn to build machine learning (ML) algorithms from scratch with the AI Theory and Coding learning path in CodeSignal Learn.
Experimenting with a variety of pre-trained models for text, image, audio, and video generation
Between the time this article was written and when you’re reading it, there will surely be new models and innovations. The important thing is that you practice your prompt engineering skills on the latest models, because that’s most likely what you’ll end up working on. Better yet, try different ones, such as OpenAI’s GPT-4o, Google’s Gemini, or Anthropic’s Claude.
One trend we are seeing that will surely continue is a shift from single-mode AIs (taking in text or image only) to multimodal ones (processing text, image, audio, and video simultaneously). Accordingly, prompt engineering doesn’t just have to work with text output: you may find that you enjoy and are good at prompting for other formats.
Pursuing a formal education in AI
Not surprisingly, AI is booming in popularity as a subject at colleges and universities. Only a few offer AI-specific degrees, but we can expect that to change.
The classes that teach you to learn how to create and work with AI models will typically be found within the computer science department. Other parts of academia, from art to medicine, are increasingly engaging with and teaching about AI from an applied perspective: how to use a model someone else has built. Still others, such as philosophy and law, consider the societal and even moral implications of AI.
What to study for a career in prompt engineering?
Computer science is the most directly relevant degree for anyone interested in building and guiding AI systems. That said, prompt engineering also requires a humanistic side. Courses in writing, languages, and linguistics will help you learn the nuances of natural language. Studying philosophy, law, psychology, and sociology will help attune your senses to bias and other unwanted output that requires a human eye to discern.
Earning prompt engineering credentials
As of the writing of this article, there appear to be no prompt engineering higher education degrees, though there are some courses that offer certificates. Generally, prompt engineering is covered as a part of a broader certification or degree in AI.
Building a strong project portfolio that showcases your skills
There are three directions you could go:
- If you can demonstrate your prompt engineering skills through your current job, that’s a great way to be useful to your employer as well as showing your capabilities to future ones. Just make sure that whatever you make public doesn’t use or expose proprietary or confidential information.
- You could also base projects around something that deeply interests you. This holds two advantages: you’ll probably do a really good job because of your topical expertise, and you’re more likely to see the project through to launch if the topic motivates you.
- One more direction is something closer to the applications that might be hiring. After all, a well-prompted bot that helps people sort out insurance questions may come across as more professional than one that holds deep conversations about, say, organic gardening. Just make sure to have someone who’s actually well-informed about the topic review your output for accuracy and usefulness.
After completing the project itself, be sure to include a narrative that explains how you crafted the prompts.
Attending industry events and networking with professional prompt engineers
As prompt engineering is a nascent field, the surrounding community is also very young. The Prompt Engineering conference in October 2024 claims to be the world’s first such conference. There are also several global and locally-based interest groups and a subreddit.
You’ll find much more opportunity with interest groups and events with a broader focus, such as gen AI or NLP. There are AI conferences for a wide variety of fields, geographies, and professions, and meetups in cities big and even not-so-big. Plus, given the huge interest in AI these days, most general tech or computing meetups will inevitably end up covering the topic.
As you go about building a network, avoid limiting yourself to people with a prompt engineer title, because, frankly, not many people have it yet. Instead, look for the types of people you’d be likely to work with in the role: AI engineers, product designers, and even policy and legal folks in the space.
Frequently asked questions
How can I start a career in prompt engineering?
Begin by filling in knowledge gaps. Start by learning and practicing the fundamentals of prompt engineering, which you can do with CodeSignal’s learning path. As you hone your prompting skills in your current job or as a side hobby, learn some Python as it’s a very common language for manipulating AI, and it’s required on many job listings. You don’t need to be a Python pro to start applying, but it will help your chances to learn enough to be able to write basic programs and make sense and apply code others (or an AI!) have written.
How long does it typically take to become a prompt engineer?
It depends on the skills you already have. If you already know how to code, you can learn the fundamentals of prompt engineering within hours and deepen your expertise in days. If you don’t already know Python or any other programming language, you can expect it to take about two months to achieve basic proficiency.
Can you pursue a career in prompt engineering through online courses?
Absolutely. The basics of prompt engineering don’t take too long to learn, and you can start an online course such as the one offered by CodeSignal right away. Build a foundation in Python with online courses, too. Once you’ve gotten the education, the next step is practice, and you can do that on your own—or, you can even ask an AI like OpenAI to critique your work.
What is the typical salary range for a prompt engineer?
A May 2024 search for US-based jobs revealed a range from $40/hour to well over $200,000 annually. The best-paying ones are at well-funded companies and require significant demonstrated coding skill and experience with AI. The lower-paying ones often have lower technical requirements..
Accelerate your career in prompt engineering with CodeSignal Learn
Prompt engineering is a brand new field that is co-evolving with generative AI. It’s a promising entry point for creatively-minded people to leverage their skills and interests to the fast-growing field of AI engineering.
You can learn the basics of prompt engineering through CodeSignal Learn without knowing a lick of code, but if you decide to continue past this initial course we strongly recommend learning the fundamentals of Python. Our courses are available for free—including limited access to our own AI-powered tutor—and you can start right away. So why not start now? We’ll see you there.