Interview series: Prompt Engineering at AI Startup
Together with Varvara, Head of Content at an AI startup, we unveil details about prompt engineering as a profession and discuss essential skills needed to succeed in it
In today’s post, we kick off a series of interviews with product-oriented professionals to explore their career trajectories in product management and tech, the exciting roles they’ve recently undertaken, and who are a part of big and small revolutions in the big tech world.
Our guest today is Varvara Reznikava, Head of Content at AI startup developing a mobile app for communicating with multiple AI characters, including digital clones of celebrities. In contrast to traditional content approaches, Varvara is pioneering investments in prompt engineering. With the surging demand for specialists who can effectively design tasks for chatbots to achieve specific outcomes, we invited Varvara to shed light on this emerging profession.
What is prompt engineering?
In short, prompt engineering is the process of crafting and refining prompts to get the best results from AI models. A prompt is essentially an instruction (a statement, set of words, questions, etc.) you send to an AI model to receive accurate and relevant responses.
Imagine you need a nice sports outfit for the gym and ask friends for help. You might say, "Hey, if you pass store X, could you get me a black t-shirt and sport shorts, size M/L? Nothing too tight, and send me pictures and prices."
Now, imagine asking someone who doesn't know you and can't send pictures. You'd need to be more specific: "Buy me a sports t-shirt and shorts from store X on George Avenue 5 or Princess Street 10. Brand: Adidas, or Nike, or Under Armour, or New Balance. Size: M/L. Shorts should be knee-length; t-shirt with short sleeves. Fabric: bamboo, or polyester, or Lyocell, or a mix. No cotton please. Total cost under 150 euros."
This detailed and clear instruction is similar to a prompt.
What skills are essential for this position?
You can be a self-learner, but to excel in this fast-growing field, you need to be inquisitive. You’ll also need:
Knowledge of AI and NLP (natural language processing): you should understand AI models, prompts, tokens, weights, etc.
A habit of continuously testing and tweaking prompts
Problem-solving skills for complex challenges
Creativity in both: thought and writing
What role does creativity play, and how do you balance creativity with technical accuracy?
Creativity is essential in prompt engineering. You need to write clearly to convey your message effectively and think creatively to solve problems when a prompt doesn’t yield the desired results. This means experimenting with different approaches until you find one that works. The technical aspect of the job can vary (depending on the role), but having a basic knowledge of NLP and AI is fundamental.
Which product or technology has excited or impressed you recently?
As expected, OpenAI's GPT-4o provided an excellent experience. Given my visual-oriented work, Kaiber remains my top choice for video production (and I'm awaiting Sora's arrival), and the recent Midjourney have produced images that are incredibly realistic.
Can you recommend a book, article, or podcast that has influenced you recently?
One of the books I’ve read lately is Difficult Conversations, which offers tips on navigating tough talks. Even though we live in a world where technology is rapidly advancing, good communication is just as important as ever. I’m also really into true crime podcasts. Listening to them is my way to unwind, but they also help me develop the skill of viewing situations from different perspectives and discovering unexpected solutions.