AI?

1. What is Artificial Intelligence?
Answer:
Artificial Intelligence (AI) is the field of computer science focused on creating systems that can perform tasks that normally require human intelligence, such as learning, reasoning, problem-solving, understanding language, recognizing images, and making decisions.

2. What is the difference between AI, Machine Learning, and Deep Learning?
Answer:

AI: The broad concept of machines performing intelligent tasks.
Machine Learning (ML): A subset of AI where systems learn patterns from data instead of being explicitly programmed.
Deep Learning (DL): A subset of ML that uses multi-layer neural networks to learn complex patterns.
Example:
AI → Self-driving car
ML → Predicting whether a customer will leave
Deep Learning → Recognizing objects from camera images

3. What is Machine Learning?
Answer:
Machine Learning is a technique that allows computers to learn from data and make predictions or decisions without being explicitly programmed for every situation.

4. What are the main types of Machine Learning?
Answer:
There are three major types:

Supervised Learning – Learns from labeled data.
Example: Email spam classification.
Unsupervised Learning – Finds patterns in unlabeled data.
Example: Customer segmentation.
Reinforcement Learning – Learns through rewards and penalties.
Example: Game-playing AI or robotics.
5. What is supervised learning?
Answer:
Supervised learning uses a dataset where the input data and the correct output are already known. The model learns the relationship between them and then uses that knowledge to predict outputs for new data.

Examples include linear regression, logistic regression, decision trees, and neural networks.

6. What is overfitting?
Answer:
Overfitting occurs when a model learns the training data too closely, including its noise and random variations. As a result, it performs very well on training data but poorly on new, unseen data.

Ways to reduce overfitting:

Use more training data
Regularization
Cross-validation
Dropout for neural networks
Reduce model complexity
Early stopping
7. What is underfitting?
Answer:
Underfitting occurs when a model is too simple to capture the important patterns in the data. It performs poorly on both training and test data.

8. What is a neural network?
Answer:
A neural network is a machine-learning model inspired by the structure of the human brain. It consists of interconnected nodes organized into layers. Neural networks learn by adjusting weights based on the difference between predicted and actual results.

9. What is an activation function?
Answer:
An activation function determines whether and how strongly a neuron should be activated. It also introduces non-linearity into a neural network.

Common activation functions include:

ReLU
Sigmoid
Tanh
Softmax
10. What is NLP?
Answer:
NLP, or Natural Language Processing, is a branch of AI that enables computers to understand, process, and generate human language.

Examples include:

Chatbots
Machine translation
Sentiment analysis
Speech recognition
Text summarization
11. What is Generative AI?
Answer:
Generative AI refers to AI systems that can create new content based on patterns learned from existing data. It can generate text, images, audio, video, and code.

Examples include large language models and image-generation models.

12. What is a Large Language Model (LLM)?
Answer:
An LLM is a machine-learning model trained on large amounts of text to understand and generate human-like language. Modern LLMs are generally based on the Transformer architecture.

13. What is a Transformer?
Answer:
A Transformer is a neural-network architecture designed to process sequences efficiently using mechanisms such as self-attention. Transformers are the foundation of many modern language models because they can capture relationships between words or tokens even when they are far apart in a sequence.

14. What is an AI hallucination?
Answer:
An AI hallucination occurs when an AI system generates information that sounds plausible but is incorrect, unsupported, or fabricated.

Ways to reduce hallucinations include:

Using reliable retrieval sources
Providing better prompts and context
Retrieval-Augmented Generation (RAG)
Fact-checking outputs
Using structured or verified data
15. What is RAG?
Answer:
RAG stands for Retrieval-Augmented Generation. It combines information retrieval with a generative AI model.

The system first retrieves relevant information from a knowledge base and then provides that information to the language model to help generate a more accurate response.

16. What is prompt engineering?
Answer:
Prompt engineering is the process of designing effective instructions or prompts so that an AI model produces more accurate, relevant, and consistent results.

17. What is the difference between classification and regression?
Answer:
Classification predicts a category or class.

Example:
Spam vs. Not Spam

Regression predicts a continuous numerical value.

Example:
Predicting the price of a house.

18. What is a confusion matrix?
Answer:
A confusion matrix is used to evaluate classification models. It contains:

True Positive (TP)
True Negative (TN)
False Positive (FP)
False Negative (FN)
It helps calculate metrics such as accuracy, precision, recall, and F1-score.

19. What is precision vs. recall?
Answer:

Precision: Of everything the model predicted as positive, how many were actually positive?

Recall: Of everything that was actually positive, how many did the model correctly identify?

For example, in fraud detection, recall may be especially important because missing fraudulent transactions can be 
﻿
High-priority interview questions
Q1. Explain an ML project you've worked on.
A strong answer should cover:

Problem → Data → Preprocessing → Features → Model selection → Training → Evaluation → Deployment → Business impact.

Q2. How do you handle an imbalanced dataset?
Possible approaches:

Class weights
Oversampling/undersampling
SMOTE
Appropriate evaluation metrics
Threshold tuning
Q3. How do you choose between different ML models?

“I first establish a simple baseline, then compare models based on the problem, dataset size, interpretability requirements, computational cost, and validation performance. I use cross-validation where appropriate and select the model based on the metric that reflects the business objective.”

Q4. Explain bias vs. variance.

High bias: model is too simple → underfitting
High variance: model is too sensitive to training data → overfitting
The goal is to find a good balance between the two.

Q5. What happens during backpropagation?

“The model calculates the loss between its prediction and the target, computes gradients of that loss with respect to the model parameters using the chain rule, and then an optimizer such as gradient descent updates the weights to reduce the loss.”

Q6. Why are Transformers important?

“Transformers use attention mechanisms to model relationships between tokens and can process sequences efficiently in parallel. They became the foundation for many modern language models and have also been applied successfully to vision, audio, and multimodal tasks.”

Q7. Explain RAG in an interview.

“RAG, or Retrieval-Augmented Generation, combines a retrieval system with a generative model. When a user asks a question, relevant documents are retrieved from a knowledge base and supplied as context to the LLM. This can improve factual grounding and allow the system to use updated or private information without retraining the entire model.”

Q8. RAG vs. fine-tuning — when would you use each?

RAG	Fine-tuning
Need current/private knowledge	Need behavior/style changes
Knowledge changes frequently	Task pattern is stable
Want source-grounded answers	Want specialized model behavior
Easier to update knowledge	Requires training process
﻿
﻿
﻿
﻿