#optimization

9 optimization interview questions, asked at Meta, Uber, Amazon and 6 more companies. Last asked Sep 2026.

+2
Linear regression with hand-written gradient descentCodingFit a one-feature linear regression by gradient descent on mean squared error, without an ML library. Then discuss choosing the learning rate, when to stop, and how to spot a…
Last asked Sep 2026
Card game: pick cards that sum to 15CodingYou get a small existing codebase for a card game played with a 36-card deck, and you may use an AI assistant. The game revolves around picking cards whose values add up to 15…
Last asked Sep 2026
Adam, AdamW and decoupled weight decayML designWrite Adam's update: the first and second moment estimates, bias correction and the step. Why is adding an L2 penalty to the loss not the same as weight decay once Adam's…
Last asked Sep 2026
Restart strategy that cuts a solver's expected run timeCodingA model solves math problems, and the time a run needs to succeed follows a known, long-tailed distribution. You may stop a run and restart it at any time. Find a restart policy…
Last asked May 2026
Find the shortest queue with few callsCodingYou can learn the length of each of several queues only through a call that is expensive. Find the shortest queue while making as few calls as you can, and explain the trade-off…
Last asked Apr 2026
Deep learning fundamentals: optimization, drift and calibrationML designAnswer conceptual questions on training: gradient descent versus SGD, convexity, sharp versus flat minima, the bias-variance trade-off, calibration and model drift, and how you…
Last asked Feb 2026
Derive the least-squares solution for linear regressionML designDerive the closed-form solution of ordinary least-squares linear regression from the loss function, and explain each step. Then discuss when the closed form is a bad idea (cost…
Last asked Feb 2026
Place K pickup points to minimize total L1 distanceCodingGiven the coordinates of N people, choose K shuttle pickup locations so that the sum of each person's Manhattan (L1) distance to their nearest pickup is as small as possible…
Last asked Dec 2025
Kernel-optimization take-home scored by speedupCodingA two-hour take-home that replaces the coding screen. AI tools are allowed. You get a simulator for a small accelerator and a working but slow kernel; your score is the speedup in…
Last asked Dec 2025