Floworks Prompt Engineer Interview: Rounds, Questions and Preparation

What the job involves

Prompt Engineers in AI and deeptech design, test, and refine prompts that make large language models produce accurate, safe, and useful outputs for real products. Day to day, they work with product and engineering teams to turn vague requirements into structured prompts, evaluate model responses, and iterate based on data and user feedback. At fast-growing companies like Floworks, they often own prompt quality for agentic workflows, tool use, and domain-specific tasks.

Role and technical knowledge

1. 1

   What is the difference between zero-shot, one-shot, and few-shot prompting, and when would you use each?

   Define each clearly and give a practical example where few-shot helps with formatting or domain jargon, while zero-shot is enough for simple tasks.

2. 2

   How would you design a prompt to reduce hallucinations in a customer support chatbot for a SaaS product?

   Mention grounding in retrieved documents, asking the model to cite sources, adding a fallback like 'I do not know', and testing with adversarial queries.

3. 3

   Explain chain-of-thought prompting. When can it hurt performance or cost?

   Describe step-by-step reasoning benefits for math and logic, but note increased token usage, latency, and possible overthinking on simple tasks.

4. 4

   What is retrieval-augmented generation (RAG), and how does prompt design change when using RAG?

   Cover how retrieved context is inserted, the need to instruct the model to rely on context, and handling irrelevant or conflicting retrieved chunks.

Practical scenarios and problem solving

1. 5

   A prompt works well in testing but fails on 20 percent of real user queries. How do you debug and improve it?

   Suggest collecting failing examples, clustering error types, adding few-shot examples for edge cases, and setting up a regression test set before and after changes.

2. 6

   You need the model to call an external API only when the user asks for current data. How would you prompt and structure this?

   Explain function calling or tool-use schemas, clear instructions on when to call, and fallback behavior when the tool returns an error.

3. 7

   How would you evaluate whether one prompt version is better than another for a summarization feature?

   Propose a mix of automatic metrics like ROUGE or BERTScore, human ratings for coherence and factuality, and a blind A/B test on a fixed dataset.

4. 8

   Your LLM feature is too slow and expensive. What prompt-level changes could help without hurting quality much?

   Discuss shortening prompts, removing redundant instructions, using smaller models for simple steps, caching, and batching or streaming where possible.

Behavioural and teamwork

1. 9

   Tell me about a time you had to explain a technical idea to someone non-technical. How did you make it clear?

   Use a specific project example, describe the analogy or visual you used, and mention the outcome or feedback you received.

2. 10

   Describe a situation where you received critical feedback on your work. What did you do?

   Show openness, a concrete change you made, and how it improved the final result or your skills.

3. 11

   How do you prioritise when you have multiple prompt experiments to run and limited time?

   Mention impact versus effort, aligning with product goals, quick wins versus long-term bets, and communicating trade-offs to the team.

4. 12

   Give an example of a time you worked in a team to meet a tight deadline. What was your role?

   Highlight clear ownership, communication, helping teammates, and what you learned about working under pressure.

HR and motivation

1. 13

   Why do you want to work as a Prompt Engineer at a fast-growing AI startup rather than a large company?

   Show comfort with ambiguity, rapid iteration, broad ownership, and genuine interest in shipping AI products quickly.

2. 14

   How do you keep up with new LLM models, tools, and prompting techniques?

   Name specific sources like research papers, model release notes, community forums, or personal projects, and mention how you test new ideas.

3. 15

   Where do you see yourself in two years in this role?

   Align with growing into deeper evaluation, agent design, or leading prompt quality for a product area, while staying hands-on.

4. 16

   What is your expected salary range, and are you open to relocating or working from an office?

   Research typical fresher ranges for AI roles in India, give a flexible range, and be honest about location preferences and notice period.