ARTIFICIAL INTELLIGENCE-LAB QUESTION

Remark/notes
QUESTION

AI-driven
You are working as an AI engineer for Myntra platform that needs to handle both recommendation
explicit user feedback (ratings) and implicit user behavior (clicks, purchases). How would system for Myntra
1 you integrate both explicit and implicit feedback into the recommendation model? What
Recommendation
hybrid techniques or algorithms could you use to improve recommendation accuracy?
Systems Engineer

Your company want reduce the customer support cost. So as an AI Engineer your Customer support
suggestion to reduce cost is to introduce chatbot customer support queries that must chatbot handling
2 handle structured queries (e.g., checking order status) and unstructured queries (e.g., structured and
general product inquiries). How would you design the architecture of the chatbot to handle unstructured queries
both types of queries effectively? Would you use a rule-based system for structured queries
and a neural network-based approach for unstructured queries, or another method? Conversational AI
Engineer
Sam want to deploy an AI model in an autonomous vehicle that must make real-time Autonomous
decisions in a complex environment with varying weather conditions, road layouts, and Systems Engineer
3 traffic patterns. What strategies would you use to ensure the AI model can generalize well
across these varying conditions? How would you handle edge cases, such as unpredictable (Develops AI models
behavior from other drivers? for real-time decision-
making under
variable road and
weather conditions.)
Due to increasing fraud in financial transaction (bank/upi/spam mail/link). What are the steps AI Fraud Detection
you can take as AI Fraud detection Specialist. Describe the evaluation metrices would you Specialist
4 prioritize for the same.
HINT – false negative, false positive Handles class
imbalance issues in
financial fraud
detection while
optimizing precision
and recall.
As a researcher you are working on a deep learning model for medical image Medical AI
segmentation, but the dataset is limited, and acquiring additional labelled data is Researcher
5 expensive and time-consuming. How would you approach training a model with a small
dataset? Would you use techniques like data augmentation, transfer learning, or few-shot Applies transfer
learning, and how would you evaluate model performance in this context? learning, data
augmentation, and
few-shot learning to
train models with
limited labeled data.
Ethical AI Specialist
Your previous hired team lacks team work which slows company progress.so, you want
Focuses on bias
to create a system that shortlisting on hiring process is transparent and also judges
6 detection, fairness,
candidates based on ATS score, hobbies and relevant team experiences. How would you
and transparency in
ensure that the AI system is fair and unbiased in selecting candidates? What techniques would
AI-based hiring
you use to prevent the model from perpetuating biases based on gender, race, or background?
models.

AI in Education
As an educational specialist you observe every student have different learning pace and Specialist
7 learning methods. So, you want to hire a team/person to create a personalized learning
Designs adaptive
experience for every student for the same. How the team/person will design the model to
learning models using
effectively personalize the learning experience? What data should be collected from students
student performance
to train the system, and how to ensure that changes keep students engaged and motivated?
data to optimize
engagement.
 Develops NLP models
You are working on an AI-based mental health assistant that interacts with users to to detect emotional
assess their emotional well-being and offer advice. The system needs to respond states and generate
empathetically to users while providing helpful insights. How would you ensure that the AI empathetic responses.
8 assistant can understand and respond appropriately to different emotional states? What natural
language processing (NLP) techniques or models would you use to recognize sentiment, and
how would you address potential challenges in maintaining empathy?

AI Ethics and
As a CEO of InstaReady you want engagement of your users. The platform should Content
recommend user various types of content, including images, videos, and text. The system Recommendation
9 needs to prioritize content that fosters positive social interactions. How would you design Engineer
the recommendation algorithm to prioritize content that promotes positive engagement and
Designs algorithms to
reduces the spread of harmful or divisive content? What methods would you use to measure
boost positive
the impact of content on user well-being?
engagement while
reducing divisive
content.

AI Content
You want to develop AI system that detect and prevent online harassment or bullying in Moderation
a social media application. The system needs to flag inappropriate content in real time. Specialist
10 What approaches would you take to accurately detect online harassment while minimizing
Develops NLP and
false positives? How would you balance the need for real-time intervention with respect for
CV models for
users' freedom of expression?
detecting and
minimizing online
harassment while
reducing false
positives.
AI in Healthcare &
You want to design a chatbot for a mental health platform that provides emotional Ethics Specialist
support and resources to users. The chatbot must detect signs of distress and respond Ensures privacy-
empathetically while ensuring user privacy. How would you design the chatbot to identify preserving NLP models
11 users in distress while maintaining their privacy? What strategies would you use to ensure that provide
empathetic responses without crossing ethical boundaries or providing misleading advice? responsible and ethical
mental health support.