Everything in AI & ML, page 9
Analyzing skewed revenue-per-user experiments
Heavy tails inflate variance and slow significance, and the mean is dominated by whales; mitigate via winsorization or capping, log transforms, CUPED variance reduction, or bootstrap and rank tests.
Interpreting a black-box gradient boosting model
Global tools like permutation importance or aggregated SHAP rank overall feature influence; local tools like per-instance SHAP or LIME explain one prediction; SHAP unifies both via additive…
Random Forest versus Gradient Boosting
Random Forest trains deep trees in parallel and averages to cut variance; boosting builds shallow trees sequentially, each correcting prior errors to cut bias, often higher accuracy but…
Stemming versus lemmatization in text preprocessing
Stemming chops affixes fast but crudely, yielding non-words; lemmatization maps to real dictionary base forms using POS, slower but accurate; skip both for embedding or transformer models.
Handling missing numerical values
Dropping rows is simple but loses data and can bias if missingness is non-random; mean or median imputation keeps rows but shrinks variance and ignores correlations; model-based imputation is…
Pandas loc versus iloc indexing
Loc selects by label and is inclusive of both endpoints; iloc selects by integer position and is exclusive of the stop; passing a string label to iloc fails.
MLE versus MAP estimation and the role of priors
MLE maximizes likelihood alone; MAP maximizes likelihood times a prior, acting as regularization that shrinks toward prior beliefs; with abundant data they converge.
Eigenvalues, eigenvectors, and their role in PCA
An eigenvector keeps direction under a matrix, its eigenvalue scales it; PCA finds eigenvectors of the covariance matrix as principal axes.
How gradient descent and the learning rate work
Gradient descent steps downhill along the negative gradient to minimize cost; the learning rate sets step size; too high diverges or oscillates, too low converges painfully slowly.
Framing ad-load tradeoffs: revenue versus retention
Define revenue plus guardrail engagement metrics, run a long-enough experiment to see retention effects, and weigh short-term lift against lifetime-value erosion.
Does forcing profile completion cause retention?
Name the confounder (engaged users self-select into completing profiles), warn that forcing it may not transfer the effect, and propose a randomized experiment.
Python Virtual Environments
A virtual environment is an isolated Python installation with its own packages, so each project gets the exact dependency versions it needs without conflicting with other projects or the system Python.
Transformer Architecture
The Transformer replaces recurrence with self-attention, letting every token directly attend to every other token in parallel. This enables long-range context and fast training on GPUs, making it the backbone of modern large language models and much of…
Generative Adversarial Network (GAN)
A GAN trains two networks in competition: a generator that fabricates fake samples and a discriminator that judges real versus fake. Their adversarial game pushes the generator toward realistic outputs, enabling image synthesis and data generation without…
Design an active learning loop for detection
Seed-train, score the pool by uncertainty plus diversity, batch to annotators, retrain, repeat.
What does N-way K-shot classification mean?
N is classes per episode, K is labeled examples per class in the support set, prediction is on a separate query set.
Filter-based vs optimization-based SLAM
EKF folds past poses into one Gaussian; optimization keeps a sparse graph and re-linearizes; the latter wins on accuracy and loop closure.
Design real-time multi-object tracking for AV
Fast detector, Kalman motion model, Hungarian association on IoU plus appearance, track lifecycle for occlusions.
Loss functions for imbalanced medical segmentation
Cross-entropy is swamped by background; Dice optimizes overlap directly; Focal down-weights easy pixels.
How is IoU computed and why prefer mIoU?
IoU is intersection over union of predicted and true pixels; mIoU averages per class; pixel accuracy is dominated by background.
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