From 71358a833ad3353701f11cc53844727e259727b6 Mon Sep 17 00:00:00 2001 From: Anurag Bhattacharjee <105806479+AnuragB2004@users.noreply.github.com> Date: Thu, 19 Jun 2025 01:52:06 +0530 Subject: [PATCH 1/2] Added GenAI interview prompts under genai/genai_interview_prompts.md --- genai/genai_interview_prompts.md | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) create mode 100644 genai/genai_interview_prompts.md diff --git a/genai/genai_interview_prompts.md b/genai/genai_interview_prompts.md new file mode 100644 index 0000000..4c04a06 --- /dev/null +++ b/genai/genai_interview_prompts.md @@ -0,0 +1,21 @@ +# Generative AI Interview Prompts + +## Prompt 1: RAG Pipeline Design + +Design a Retrieval-Augmented Generation (RAG) system that can query a company’s internal documentation (PDFs, Notion, internal wiki) to generate real-time customer responses. + +## Prompt 2: LLM Evaluation Metrics + +You are tasked with evaluating three LLMs (e.g., GPT-4, Gemini, Claude). Define objective metrics and testing strategies for accuracy, hallucination, and latency in a QA chatbot. + +## Prompt 3: Prompt Injection Mitigation + +Explain how prompt injection attacks work. Propose a defense mechanism to secure your GenAI app used by thousands of users daily. + +## Prompt 4: Fine-Tuning vs RAG + +Compare and contrast fine-tuning an LLM with using a RAG system. In what scenarios would one be preferred over the other? + +## Prompt 5: Multimodal GenAI Product Design + +Design a GenAI tool that accepts images and voice inputs from doctors and generates medical notes. What models, pipelines, and data formats would you use? \ No newline at end of file From 83ab65d317404c60c372c874dc654de09ef2d7a9 Mon Sep 17 00:00:00 2001 From: Anurag Bhattacharjee Date: Thu, 19 Jun 2025 02:06:28 +0530 Subject: [PATCH 2/2] Added GenAI interview prompts under genai/genai_interview_prompts.md Signed-off-by: Anurag Bhattacharjee --- genai/genai_interview_prompts.md | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/genai/genai_interview_prompts.md b/genai/genai_interview_prompts.md index 4c04a06..8c038ae 100644 --- a/genai/genai_interview_prompts.md +++ b/genai/genai_interview_prompts.md @@ -1,21 +1,21 @@ # Generative AI Interview Prompts -## Prompt 1: RAG Pipeline Design +## Prompt 1: RAG System Design -Design a Retrieval-Augmented Generation (RAG) system that can query a company’s internal documentation (PDFs, Notion, internal wiki) to generate real-time customer responses. +Design a RAG (Retrieval-Augmented Generation) pipeline for a chatbot that uses internal documentation from PDFs and Notion pages. -## Prompt 2: LLM Evaluation Metrics +## Prompt 2: LLM Evaluation -You are tasked with evaluating three LLMs (e.g., GPT-4, Gemini, Claude). Define objective metrics and testing strategies for accuracy, hallucination, and latency in a QA chatbot. +How would you benchmark GPT-4 vs Claude vs Gemini for factual accuracy, speed, and context understanding? -## Prompt 3: Prompt Injection Mitigation +## Prompt 3: Prompt Injection -Explain how prompt injection attacks work. Propose a defense mechanism to secure your GenAI app used by thousands of users daily. +Explain prompt injection in LLMs. How would you protect an app like ChatGPT from it? -## Prompt 4: Fine-Tuning vs RAG +## Prompt 4: Fine-tuning vs RAG -Compare and contrast fine-tuning an LLM with using a RAG system. In what scenarios would one be preferred over the other? +Compare fine-tuning vs using a RAG approach. Which is better for enterprise Q&A systems? -## Prompt 5: Multimodal GenAI Product Design +## Prompt 5: Multimodal GenAI Design -Design a GenAI tool that accepts images and voice inputs from doctors and generates medical notes. What models, pipelines, and data formats would you use? \ No newline at end of file +Design a GenAI tool where doctors speak symptoms and it generates medical summaries. What models and architecture will you use? \ No newline at end of file