AI Engineer Interview Questions
A comprehensive guide covering Core ML, Deep Learning, LLMs, NLP, System Design, MLOps, and Computer Vision for AI and Machine Learning roles.
Total Questions:150
Difficulty Levels:
BeginnerIntermediateAdvanced
0%
Overall Progress
0/150
Status
Problem
Level
2.What's the difference between L1 and L2 regularization and when do you use each?Medium
2.What's the difference between L1 and L2 regularization and when do you use each?
Medium
3.Walk me through how backpropagation works.Hard
3.Walk me through how backpropagation works.
Hard
4.What is overfitting and how do you prevent it in real projects?Easy
4.What is overfitting and how do you prevent it in real projects?
Easy
5.Explain precision vs recall - when do you optimize for which?Medium
5.Explain precision vs recall - when do you optimize for which?
Medium
6.What's the bias-variance tradeoff?Medium
6.What's the bias-variance tradeoff?
Medium
7.How do you handle imbalanced datasets?Medium
7.How do you handle imbalanced datasets?
Medium
8.What's cross-validation and why do we use it?Easy
8.What's cross-validation and why do we use it?
Easy
9.Explain the difference between bagging and boosting.Medium
9.Explain the difference between bagging and boosting.
Medium
10.When would you use a decision tree vs neural network?Medium
10.When would you use a decision tree vs neural network?
Medium
11.Explain how a neural network learns.Easy
11.Explain how a neural network learns.
Easy
12.What activation functions do you know and when do you use each?Medium
12.What activation functions do you know and when do you use each?
Medium
13.Why is ReLU better than sigmoid?Medium
13.Why is ReLU better than sigmoid?
Medium
14.What's batch normalization and why use it?Hard
14.What's batch normalization and why use it?
Hard
15.What's dropout and how does it work?Medium
15.What's dropout and how does it work?
Medium
16.Explain CNN architecture - what are convolution and pooling layers?Medium
16.Explain CNN architecture - what are convolution and pooling layers?
Medium
17.What's the difference between RNN, LSTM, and GRU?Hard
17.What's the difference between RNN, LSTM, and GRU?
Hard
18.Explain the Transformer architecture.Hard
18.Explain the Transformer architecture.
Hard
19.What's attention mechanism?Hard
19.What's attention mechanism?
Hard
20.What's the vanishing gradient problem?Hard
20.What's the vanishing gradient problem?
Hard
21.What's the difference between GPT and BERT?Medium
21.What's the difference between GPT and BERT?
Medium
22.Explain how GPT works at a high level.Medium
22.Explain how GPT works at a high level.
Medium
23.What's fine-tuning vs prompt engineering?Easy
23.What's fine-tuning vs prompt engineering?
Easy
24.What's RAG (Retrieval Augmented Generation)?Hard
24.What's RAG (Retrieval Augmented Generation)?
Hard
25.How would you build a chatbot using LLMs?Medium
25.How would you build a chatbot using LLMs?
Medium
26.What's the difference between zero-shot, one-shot, and few-shot learning?Easy
26.What's the difference between zero-shot, one-shot, and few-shot learning?
Easy
27.What's temperature in LLM sampling?Easy
27.What's temperature in LLM sampling?
Easy
28.How do you prevent hallucinations in LLMs?Medium
28.How do you prevent hallucinations in LLMs?
Medium
29.What's vector database and when do you need it?Medium
29.What's vector database and when do you need it?
Medium
30.What's embedding and how is it used?Easy
30.What's embedding and how is it used?
Easy
31.Explain tokenization.Easy
31.Explain tokenization.
Easy
32.What's RLHF (Reinforcement Learning from Human Feedback)?Hard
32.What's RLHF (Reinforcement Learning from Human Feedback)?
Hard
33.How would you evaluate an LLM's performance?Medium
33.How would you evaluate an LLM's performance?
Medium
34.What's prompt injection?Medium
34.What's prompt injection?
Medium
35.How would you reduce costs when using LLM APIs?Medium
35.How would you reduce costs when using LLM APIs?
Medium
36.Walk me through your ML project from data to deployment.Easy
36.Walk me through your ML project from data to deployment.
Easy
37.How do you split your data and why?Easy
37.How do you split your data and why?
Easy
38.What's your feature engineering process?Medium
38.What's your feature engineering process?
Medium
39.How do you handle missing data?Easy
39.How do you handle missing data?
Easy
40.How do you handle categorical variables?Easy
40.How do you handle categorical variables?
Easy
41.What's data leakage and how do you prevent it?Medium
41.What's data leakage and how do you prevent it?
Medium
42.How do you debug a model that's not learning?Hard
42.How do you debug a model that's not learning?
Hard
43.How do you know when to stop training?Easy
43.How do you know when to stop training?
Easy
44.What metrics do you use for classification vs regression?Easy
44.What metrics do you use for classification vs regression?
Easy
45.How do you deploy an ML model to production?Medium
45.How do you deploy an ML model to production?
Medium
46.What's model drift and how do you handle it?Hard
46.What's model drift and how do you handle it?
Hard
47.How do you monitor models in production?Medium
47.How do you monitor models in production?
Medium
48.What's A/B testing for ML models?Medium
48.What's A/B testing for ML models?
Medium
49.How do you handle different versions of models?Medium
49.How do you handle different versions of models?
Medium
50.What's the difference between batch and online inference?Medium
50.What's the difference between batch and online inference?
Medium
51.Design a recommendation system for Netflix.Hard
51.Design a recommendation system for Netflix.
Hard
52.Design a fraud detection system for a bank.Hard
52.Design a fraud detection system for a bank.
Hard
53.Design a search ranking system.Hard
53.Design a search ranking system.
Hard
54.Design a real-time content moderation system.Hard
54.Design a real-time content moderation system.
Hard
55.How would you build Instagram's image search?Hard
55.How would you build Instagram's image search?
Hard
56.Design a spam detection system for email.Medium
56.Design a spam detection system for email.
Medium
57.How would you build a personalized news feed?Hard
57.How would you build a personalized news feed?
Hard
58.Design a chatbot for customer service.Medium
58.Design a chatbot for customer service.
Medium
59.How would you scale an ML model to handle millions of requests?Hard
59.How would you scale an ML model to handle millions of requests?
Hard
60.Design a system to detect fake news.Hard
60.Design a system to detect fake news.
Hard
61.Implement linear regression from scratch.Medium
61.Implement linear regression from scratch.
Medium
62.Write a function to calculate accuracy, precision, recall.Easy
62.Write a function to calculate accuracy, precision, recall.
Easy
63.Implement k-means clustering.Medium
63.Implement k-means clustering.
Medium
64.Code softmax function.Easy
64.Code softmax function.
Easy
65.Implement train-test split.Easy
65.Implement train-test split.
Easy
66.Write code for data normalization.Easy
66.Write code for data normalization.
Easy
67.Implement cross-entropy loss.Medium
67.Implement cross-entropy loss.
Medium
68.Code a simple neural network in NumPy.Hard
68.Code a simple neural network in NumPy.
Hard
69.Write data preprocessing pipeline.Medium
69.Write data preprocessing pipeline.
Medium
70.Implement gradient descent.Easy
70.Implement gradient descent.
Easy
71.Code for handling imbalanced data.Medium
71.Code for handling imbalanced data.
Medium
72.Write evaluation metrics code.Easy
72.Write evaluation metrics code.
Easy
73.Implement confusion matrix.Easy
73.Implement confusion matrix.
Easy
74.Code for model inference API.Medium
74.Code for model inference API.
Medium
75.Write unit tests for ML code.Medium
75.Write unit tests for ML code.
Medium
76.How do you build a model in PyTorch vs TensorFlow?Easy
76.How do you build a model in PyTorch vs TensorFlow?
Easy
77.What's torch.nn.Module?Easy
77.What's torch.nn.Module?
Easy
78.How do you create custom PyTorch Dataset?Medium
78.How do you create custom PyTorch Dataset?
Medium
79.What's the difference between model.train() and model.eval()?Easy
79.What's the difference between model.train() and model.eval()?
Easy
80.How do you save and load models?Easy
80.How do you save and load models?
Easy
81.Explain PyTorch autograd.Medium
81.Explain PyTorch autograd.
Medium
82.What's the difference between tensor.detach() and tensor.data?Hard
82.What's the difference between tensor.detach() and tensor.data?
Hard
83.How do you use Hugging Face Transformers?Easy
83.How do you use Hugging Face Transformers?
Easy
84.How do you fine-tune a BERT model?Medium
84.How do you fine-tune a BERT model?
Medium
85.What's tokenizer in Hugging Face?Easy
85.What's tokenizer in Hugging Face?
Easy
86.How do you containerize an ML model?Medium
86.How do you containerize an ML model?
Medium
87.What's Docker and why use it for ML?Easy
87.What's Docker and why use it for ML?
Easy
88.How do you use Kubernetes for ML?Hard
88.How do you use Kubernetes for ML?
Hard
89.What's CI/CD for ML models?Medium
89.What's CI/CD for ML models?
Medium
90.How do you track experiments?Easy
90.How do you track experiments?
Easy
91.What tools do you use for model versioning?Medium
91.What tools do you use for model versioning?
Medium
92.How do you handle feature engineering in production?Hard
92.How do you handle feature engineering in production?
Hard
93.What's the difference between training and serving infrastructure?Medium
93.What's the difference between training and serving infrastructure?
Medium
94.How do you optimize model latency?Hard
94.How do you optimize model latency?
Hard
95.How do you reduce model size for deployment?Hard
95.How do you reduce model size for deployment?
Hard
96.What's quantization?Medium
96.What's quantization?
Medium
97.How do you handle GPU memory issues?Medium
97.How do you handle GPU memory issues?
Medium
98.What's batch size and how do you choose it?Easy
98.What's batch size and how do you choose it?
Easy
99.How do you debug OOM (Out of Memory) errors?Medium
99.How do you debug OOM (Out of Memory) errors?
Medium
100.How do you profile model performance?Hard
100.How do you profile model performance?
Hard
101.How do you work with large datasets that don't fit in memory?Medium
101.How do you work with large datasets that don't fit in memory?
Medium
102.What's the difference between SQL and NoSQL for ML?Easy
102.What's the difference between SQL and NoSQL for ML?
Easy
103.How do you store training data?Medium
103.How do you store training data?
Medium
104.What's data versioning?Medium
104.What's data versioning?
Medium
105.How do you create a data pipeline?Medium
105.How do you create a data pipeline?
Medium
106.What's Apache Spark and when do you use it?Hard
106.What's Apache Spark and when do you use it?
Hard
107.How do you handle streaming data?Hard
107.How do you handle streaming data?
Hard
108.What's ETL pipeline?Easy
108.What's ETL pipeline?
Easy
109.How do you validate data quality?Medium
109.How do you validate data quality?
Medium
110.What's feature store?Hard
110.What's feature store?
Hard
111.Explain object detection vs image classification.Easy
111.Explain object detection vs image classification.
Easy
112.What's YOLO?Medium
112.What's YOLO?
Medium
113.What's the difference between semantic and instance segmentation?Medium
113.What's the difference between semantic and instance segmentation?
Medium
114.How do you handle different image sizes?Easy
114.How do you handle different image sizes?
Easy
115.What image augmentation techniques do you use?Easy
115.What image augmentation techniques do you use?
Easy
116.What's transfer learning in computer vision?Easy
116.What's transfer learning in computer vision?
Easy
117.How would you build a face recognition system?Hard
117.How would you build a face recognition system?
Hard
118.What's U-Net architecture?Hard
118.What's U-Net architecture?
Hard
119.How do you evaluate object detection models?Medium
119.How do you evaluate object detection models?
Medium
120.What's IoU (Intersection over Union)?Easy
120.What's IoU (Intersection over Union)?
Easy
121.You have a model with 95% accuracy but it's failing in production. Why?Medium
121.You have a model with 95% accuracy but it's failing in production. Why?
Medium
122.Your model works on training data but fails on test data. What do you do?Easy
122.Your model works on training data but fails on test data. What do you do?
Easy
123.You need to reduce model inference time by 50%. How?Hard
123.You need to reduce model inference time by 50%. How?
Hard
124.Your dataset has 1,000 positive and 100,000 negative examples. How do you handle this?Medium
124.Your dataset has 1,000 positive and 100,000 negative examples. How do you handle this?
Medium
125.Your model's performance is degrading over time. What's happening?Medium
125.Your model's performance is degrading over time. What's happening?
Medium
126.You need to explain your model to non-technical stakeholders. How?Easy
126.You need to explain your model to non-technical stakeholders. How?
Easy
127.Your training is taking too long. What do you optimize?Medium
127.Your training is taking too long. What do you optimize?
Medium
128.You have limited labeled data. What techniques do you use?Hard
128.You have limited labeled data. What techniques do you use?
Hard
129.Your model needs to run on mobile devices. What do you do?Medium
129.Your model needs to run on mobile devices. What do you do?
Medium
130.You're asked to improve an existing model by 5%. What's your approach?Medium
130.You're asked to improve an existing model by 5%. What's your approach?
Medium
131.Tell me about a challenging ML project you worked on.Medium
131.Tell me about a challenging ML project you worked on.
Medium
132.Describe a time when your model failed in production.Medium
132.Describe a time when your model failed in production.
Medium
133.How do you stay updated with AI research?Easy
133.How do you stay updated with AI research?
Easy
134.Walk me through your ML project from start to finish.Easy
134.Walk me through your ML project from start to finish.
Easy
135.What's the biggest mistake you made in an ML project?Medium
135.What's the biggest mistake you made in an ML project?
Medium
136.How do you prioritize features when building models?Medium
136.How do you prioritize features when building models?
Medium
137.Describe a time you had to explain ML to non-technical people.Easy
137.Describe a time you had to explain ML to non-technical people.
Easy
138.What's your approach when you get a new ML problem?Easy
138.What's your approach when you get a new ML problem?
Easy
139.How do you handle disagreements about model approach?Medium
139.How do you handle disagreements about model approach?
Medium
140.Why do you want to work in AI?Easy
140.Why do you want to work in AI?
Easy
141.What's matrix multiplication and why is it important?Easy
141.What's matrix multiplication and why is it important?
Easy
142.Explain eigenvalues and eigenvectors.Hard
142.Explain eigenvalues and eigenvectors.
Hard
143.What's probability distribution?Easy
143.What's probability distribution?
Easy
144.Explain Bayes theorem.Medium
144.Explain Bayes theorem.
Medium
145.What's derivative and why is it important in ML?Easy
145.What's derivative and why is it important in ML?
Easy
146.What's the chain rule?Medium
146.What's the chain rule?
Medium
147.Explain logarithm and its use in ML.Easy
147.Explain logarithm and its use in ML.
Easy
148.What's exponential function?Easy
148.What's exponential function?
Easy
149.What's normalization in statistics?Easy
149.What's normalization in statistics?
Easy
150.What's standard deviation and variance?Easy
150.What's standard deviation and variance?
Easy