AI Ultimate AI Engineer 2

Question 1

Which technique reduces LLM hallucinations by grounding responses in retrieved documents?

Accepted Answer

Retrieval-Augmented Generation (RAG)

Answer

RAG retrieves relevant documents at inference time and conditions the model's output on that grounded context, reducing hallucinations.

Question 2

In transformer attention, what does the 'key' vector represent?

Accepted Answer

What information the token offers for matching

Answer

Keys represent what each token offers; queries ask what is needed, and the dot product between them scores relevance.

Question 3

Which Python library is the industry standard for distributed training of large neural networks across multiple GPUs?

Accepted Answer

PyTorch + DeepSpeed/FSDP

Answer

PyTorch with DeepSpeed or FSDP (Fully Sharded Data Parallel) is the standard for large-scale distributed training.

Question 4

What is the primary purpose of a vector database like Pinecone or Weaviate in an AI system?

Accepted Answer

Enabling fast approximate nearest-neighbor search over embeddings

Answer

Vector databases index high-dimensional embeddings and support ANN search, enabling semantic retrieval at scale.

Question 5

An AI model deployed to production shows increasing latency over time without code changes. What is the most likely cause?

Accepted Answer

Memory leak in the inference server

Answer

A gradual latency increase without code changes typically points to a memory leak accumulating in the serving process over time.

AI - Engineer Practice Test