Interview questions in Data Science & Analytics, page 5
What is a word embedding and how does it beat one-hot encoding?
Tests dense semantic vectors versus sparse one-hot representations. A good answer defines embeddings as learned real-valued vectors where similar words are close, contrasts them with orthogonal one-hot vectors lacking similarity, and names Word2Vec or GloVe.
Describe Transformer architecture and why self-attention beats recurrence
This tests parallelization and long-range dependencies. A strong answer outlines the encoder-decoder stack with multi-head self-attention, contrasts O(1) sequential steps versus RNNs' O(n) unrolling, and warns that describing it as averaging misses key ideas.
How do you leverage and fine-tune BERT for niche classification?
Tests transfer learning with scarce labels. Outline: pick a domain-adjacent checkpoint, add a classification head, use learning rates near 2e-5 with early stopping, and stratify tiny validation splits.
RL components and how Q-learning works
Agent acts on the environment, observes state and reward, seeking to maximize cumulative discounted reward; Q-learning iteratively updates Q(s,a) toward reward plus discounted best…
Walk me through a CNN's layers for image classification
Tests hierarchical feature extraction in CNNs. Answer: conv filters learn edges-to-objects with shared weights, pooling reduces dimensions and adds invariance, fully-connected layers classify.
Exploration versus exploitation: epsilon-greedy and UCB
Exploit current best to earn reward, explore to discover better options; epsilon-greedy explores randomly with probability epsilon; UCB explores by an uncertainty bonus…
Why Transformers use multi-head attention
A single head averages into one representation subspace; multiple heads attend in parallel to different subspaces, letting the model capture diverse relations like syntax and coreference at once, then…
Mode collapse in GANs and how to fix it
Mode collapse is the generator producing few similar outputs, missing data diversity to fool the discriminator; fixes include Wasserstein loss, minibatch discrimination, unrolled GANs, and feature…
Spark transformations versus actions
Transformations like map and filter are lazy and build a lineage DAG returning a new RDD; actions like count or collect trigger execution and return a value to the driver.
HDFS purpose and fault tolerance
HDFS stores huge files across many commodity machines as large blocks, replicating each block across nodes for fault tolerance; unlike NTFS or ext4 it is distributed, write-once, and optimized for…
The MapReduce paradigm explained
Map applies a function to each input record emitting key-value pairs in parallel; a shuffle groups values by key; reduce aggregates each key's values into a result.
Spark broadcast join versus shuffle join
A broadcast join sends the small table to every executor so the large table joins locally with no shuffle of its rows; the default sort-merge join shuffles both tables across the network, which is costly.
Data skew in Spark and salting
Data skew is uneven key distribution sending most rows to one partition and straggler task; salting appends a random suffix to hot keys to spread them across partitions, joining in two…
Spark RDDs, DataFrames, and Datasets
RDDs are low-level typed object collections with no built-in optimization; DataFrames are named columns optimized by Catalyst and Tungsten; Datasets add compile-time type safety in…
Diagnosing Spark executor OutOfMemoryError
Check the Spark UI for skew and spills, inspect executor memory and partition count, find culprits like wide collect, huge shuffles, or skewed keys, and fix via more partitions, memory tuning, or…
Visualizing long-term trend versus seasonality
A line chart over the full three years, often with a moving average, shows the long-term trend; a seasonal plot overlaying each year by month, or a month-of-year box plot, reveals…
Reconcile rising sign-ups with falling revenue per user
Reconcile the metrics via total revenue and segment mix, frame the tradeoff, recommend an action.
Two ways accurate data can still mislead in a chart
Name distortions like truncated axes or cherry-picked ranges, give the fix for each, explain why each misleads.
Explain k-means user segments to a marketing team
Name each segment, profile its defining traits, show size and value, recommend an action.
Deploy a saved model as a REST prediction service
Load the artifact, wrap it in a predict API, containerize, host with autoscaling, add monitoring.
We are hiring for this. Every open role lists the topics its interview covers, so you can prepare for the real thing rather than guessing.
See open roles