Train a model to extract highlight snippets from textML designEach training example is a passage of text with the snippet that should be highlighted and its start and end positions. Complete a PyTorch training pipeline that learns to predict…
Last asked Oct 2026
+7
ML fundamentals deep dive for ML rolesML designAfter a short walk through your resume, expect fast questions on machine learning basics, usually centred on your own area (language models, vision, recommendations). For each…
Last asked Oct 2026
Design a work-together distance between two employeesML designDesign a distance d(u1, u2) between any two people in a company that measures how closely they work together: smaller means closer. Decide which signals to use (shared documents…
Last asked Sep 2026
+2
Transformer and attention fundamentalsML designExplain how a transformer works from the inside: how self-attention is computed, why the scores are scaled, how encoder and decoder masking differ, and what multiple heads buy…
Research engineering training discussionML designDiscuss training approaches for long-context and long-horizon tasks, then explore past projects, environment design and reward design with team leads. Reported on 1point3acres'…
Last asked Sep 2026
+2
Implement multi-head self-attentionML designWrite multi-head self-attention from scratch: project the input to queries, keys and values, split them into heads, compute scaled dot-product attention, combine the heads and…
Last asked Sep 2026
Post-train a 7B model to code better for one enterpriseML designAn enterprise customer wants a 7-billion-parameter model that writes better code for them. Design the post-training pipeline end to end: collecting and cleaning data, supervised…
Last asked Sep 2026
Make an LLM recognize questions past its training cutoffML designChat assistants sometimes answer a question about recent events by saying their knowledge stops at a certain date. You are building an LLM: how would you get it to notice that a…
Last asked Sep 2026
Defend your own research paper under hard questionsML designYou present one of your published papers to an interviewer who has already read it and may skip your slides to go straight at its weak spots. Be ready to say whether the…
Last asked Sep 2026
+1
ML design for retrieval: embeddings, hybrid search and rankingML designDesign the machine learning behind the retrieval stage of a question-answering system. Discuss how you train the embedding model, how retrieval combines dense and keyword search…
Last asked Sep 2026
Matmul performance modeling on a fixed GPUML designEstimate wall-clock time for matrix multiplications of varying shapes on a given accelerator from arithmetic intensity and memory movement, then extend the model to a multi-layer…
Last asked Sep 2026
+1
Design a harmful content detection systemML designDetect harmful posts (hate, violence, self-harm, spam) across text and images, with high recall on severe harms and a path to human review.
Last asked Sep 2026
Predict click-through rate for music recommendationsML designDesign a model that predicts whether a user will click a recommended song. Explain how you would embed songs, how you handle the different kinds of features, and how you would…
Last asked Sep 2026
How would you verify a smile detector in a photo app?ML designYou are on the Google Photos team and your feature detects smiling faces in photos. How would you check that the detection is correct? Cover what counts as a smile and who decides…
Last asked Sep 2026
Components of an LLM agent's memoryML designName the common components of an LLM agent's memory (for example the context window, a short-term scratchpad or conversation buffer, long-term stores of facts and past episodes…
Last asked Sep 2026
How data collection shapes model performanceML designPick a project from your resume and explain how its data was collected and processed. Then discuss how data choices (coverage, label quality, noise, imbalance, distribution shift)…
Last asked Sep 2026
Decoder-only LLM architecture and how vLLM serves itML designExplain how a decoder-only language model is built, from tokens through the stack of attention and feed-forward layers to predicting the next token. Then explain how an inference…
Last asked Sep 2026
+3
Design a personalized search ranking systemML designRank search results for each user's query and context so they find and book (or order) what they want, learning from clicks and conversions.
Keep GPUs fed: fix a training job that waits on its dataML designA training job's GPUs sit idle a third of the time waiting for the next batch. How do you prove the input pipeline is the bottleneck, and then fix it? Cover profiling, parallel…
Last asked Sep 2026
+2
Reward-model RLHF versus direct preference optimizationML designWalk through a classic RLHF pipeline: supervised fine-tuning, a reward model trained on preference pairs, then policy optimization (PPO) with a KL penalty to the reference model…
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
The forward and reverse processes of a diffusion modelML designExplain how a diffusion model is trained and how it generates. Describe the forward process that adds noise step by step, the reverse process the network learns, the usual…
Last asked Sep 2026
RMSNorm versus LayerNorm in a transformer blockML designWrite the formulas for LayerNorm and RMSNorm and say what each normalizes. What does RMSNorm give up by skipping the mean, why does it work about as well in practice, and why do…
Last asked Sep 2026
Mixture of experts and expert parallelismML designExplain how a mixture-of-experts transformer layer works and follow a batch of tokens through it: the router, top-k selection, the all-to-all exchange that sends tokens to experts…
Last asked Sep 2026
+1
Tensor parallelism versus pipeline parallelismML designA transformer is too big for one accelerator. Explain how tensor parallelism and pipeline parallelism each split it: what is cut, what crosses the wire and when, where the idle…
Last asked Sep 2026
Predict employee attrition and explain what drives itML designYou get an HR dataset with one row per employee (age, job role, monthly income, overtime and similar columns) and a label saying whether the person left. Build a model that…
Last asked Aug 2026
Choose storage tiers from predicted file accessML designA cloud storage service can keep each user file in a fast, expensive tier or a cheap, slow archive tier, and moving data or reading from the archive costs money. Decide which tier…
Last asked Aug 2026
Forecast storage demand across data centersML designForecast how much storage each of about 100 data centers will need, so capacity can be bought ahead of time. Cover the target and horizon (daily demand, weeks or months ahead)…
Last asked Aug 2026
Design watch-next video recommendations and derive the lossesML designDesign the "watch next" list for a large video platform: while someone watches a video, the system picks and orders the videos to offer next. This round goes deep on modeling…
Pipeline parallelism in GRPO trainingML designExplain pipeline parallelism and how it applies when a language model is trained with GRPO: how the model is split into stages across devices, how micro-batches keep the stages…
Last asked Aug 2026
KL divergence in language modelsML designDefine KL divergence and explain what it measures. Then explain where it is used with large language models and why, for example as a penalty that keeps a fine-tuned model close…
Last asked Aug 2026
What perplexity tells you about a language modelML designDefine perplexity and explain how to read it: what a lower or higher number means, how it relates to cross-entropy, how it is used for today's large language models, and where it…
Last asked Aug 2026
Importance sampling in LLM trainingML designExplain importance sampling: what it is and why it works. Then explain where it is used when training large language models and why it is used there. Expect to go beyond the…
Last asked Aug 2026
VAEs: the ELBO, the KL term and reparameterizationML designDefine a variational autoencoder and its evidence lower bound (ELBO). Why does the objective include a KL divergence term, and what does it do? Explain the reparameterization…
Last asked Aug 2026
Detect and prevent reward hackingML designExplain reward hacking when a language model is trained with reinforcement learning: what it looks like, how you would detect it, and what you would change in the reward, the data…
Last asked Aug 2026
Migrate 200 TB of media and feed it to trainingML designMove about 200 terabytes of media files to new storage and build the input pipeline that streams them to model training. Cover how you copy that much data reliably (parallelism…
Last asked Aug 2026
Roofline analysis of a BF16 MLP layerML designApply roofline analysis to an MLP layer running in BF16: count its floating-point operations and the bytes it moves, compute its arithmetic intensity, and decide whether it is…
Last asked Aug 2026
Where to apply ML in a task marketplace for driversML designA platform lets customers post tasks that drivers carry out. When a driver finishes, they upload evidence that the task was done. In a short, open discussion, walk the flow from…
Last asked Aug 2026
Design a platform to train and serve ML modelsML designDesign a platform that lets teams train machine learning models and serve them in production. Cover data access, training jobs, storing and versioning models, deployment, and…
Last asked Aug 2026
GPU and inference performance fundamentalsML designAnswer questions on GPU performance: Amdahl's law, HBM versus on-chip SRAM, matrix-multiply cost and tiling, CPU versus GPU execution, tensor and pipeline parallelism, and where…
Predict a future point with a fitted lineML designFit a straight line to a set of data points and use it to predict the value at a future point. Explain how the fit is computed and how far you would trust the prediction.
Last asked Jul 2026
Review a compact GPT-style transformer implementationML designYou are given a short implementation of a GPT-style decoder-only transformer. Read it and review it: explain what each part does (embeddings, attention with its causal mask…
Last asked Jul 2026
Design a large-scale recommendation system end to endML designDesign a recommendation system from scratch. When a user opens a surface, it picks a short, ordered list of items out of a very large catalog that changes all the time, and it…
Last asked Jul 2026
Rapid-fire questions on LLM internalsML designA fast round of short questions on how LLMs work: transformer and attention internals, training and fine-tuning choices, and how models are run in practice. Each answer should be…
Last asked Jul 2026
LLM foundations: architecture, adaptation and steeringML designA broad conversation on large language models: how the architecture works, the ways to adapt a model to a task (full fine-tuning, parameter-efficient methods, prompting), and the…
Last asked Jun 2026
Design a recommender for local sports teams and newsML designRecommend local sports teams, and the news about them, to each user. Discuss what data you collect and how it is produced, what you know about the user (location, follows…
Last asked Jun 2026
GRPO deep dive for a reasoning LLMML designExplain GRPO and why it can drop the critic that PPO uses. Then discuss the parallelism used to train a reasoning language model, multi-head latent attention (MLA), and how you…
Last asked Jun 2026
Tokenizer design and KL-regularized fine-tuningML designDiscuss how you would design a tokenizer for a language model and the trade-offs involved. Then explain supervised fine-tuning with a KL penalty that keeps the model close to its…
Last asked Jun 2026
Evaluate noisy data for LLM post-trainingML designYou have data collected for post-training a language model and some of it is noisy. How would you measure its quality, find the bad examples, and decide what to keep, fix or drop…
Last asked May 2026
Mine new information from a huge unlabeled image corpusML designYou have a very large corpus of unlabeled images. Design a system that surfaces examples unlike anything already in the training data, and that finds the images containing a given…
Last asked May 2026
+4
Design a payment fraud detection systemML designScore every card payment for fraud within tens of milliseconds, balancing blocked fraud against declined good customers, with labels that arrive weeks later.
Last asked May 2026
Design candidate search for a talent marketplaceML designDesign a system that lets a company search for candidates and ranks them for a role: how candidate profiles are represented and indexed, how a query or job description retrieves…
Last asked May 2026
Semantic search with user embeddingsML designDesign search that uses embeddings of users together with embeddings of listings, so results match both the request and the person searching. Cover how the user and item…
Mechanistic interpretability experiment on double descentML designDesign and run a small-model experiment reproducing sample-aspect double descent, explain the phenomenon from theory, and propose a remedy. The deliverable is the experiment plus…
Last asked Apr 2026
Multitask regression with targets on different scalesML designAn ML coding exercise that follows a fraud-detection design discussion. Instead of a classification label, the model has two regression targets: one takes only negative values…
Last asked Apr 2026
ML configuration system with inheritance and validationML designDesign and partly implement an ML configuration system covering schemas, inheritance and overrides, validation, and reproducibility. python file=solution.py class Field: def…
Last asked Apr 2026
+5
Design a feed recommendation systemML designRank the items in each user's home feed so they engage more, choosing from millions of candidates within a couple of hundred milliseconds.
Last asked Apr 2026
Design a real-time sensor intelligence systemML designDesign a system that takes in sensor data in real time and runs machine learning on it to produce useful signals as the data arrives. Cover ingestion, the models and how they are…
Last asked Apr 2026
Defend your research direction and an experiment designML designDiscuss your research area: what the state of the art is, the strengths and weaknesses of the main approaches, your own hands-on experience, and where you think the field is…
Last asked Apr 2026
CLIP contrastive lossML designImplement the CLIP-style contrastive loss for a batch of matching image and text embeddings. Build the image-text similarity matrix, take the cross-entropy against the diagonal…
Last asked Apr 2026
Cold start in ranking systemsML designHow do you rank new items or serve new users that have no interaction history? Discuss content features, exploration, priors and boosting, how to move from cold to warm signals…
Last asked Mar 2026
Design a drop-off spot selector for a robotaxiML designDesign an ML system that picks where an autonomous ride-hail vehicle should drop off its passenger near the destination. Cover the candidate spots, the signals and labels you…
Last asked Mar 2026
Design a nearby places recommendation systemML designDesign the ML system that recommends places to visit (restaurants, shops, attractions) to a user in a maps or local-discovery product. The user may be on the move, so location and…
Debug a GRPO training script and defend the RL mathML designBelow is a small group relative policy optimization (GRPO) trainer in plain Python. The policy is a table of logits per prompt, and the reward is the share of tokens that match…
Last asked Mar 2026
Dataset size, generalization and U-Net skip connectionsML designAnswer ML fundamentals questions: how training set size affects generalization and overfitting, how you would tell whether more data would help, and why U-Net uses skip…
Last asked Mar 2026
Deep and Cross Network v1 versus v2, and A/B testing itML designExplain how the cross network in DCN works, what changed in DCN v2 (a full weight matrix in the cross layer, low-rank versions, mixture of experts) and why it matters. Then…
Last asked Mar 2026
Design a cost-per-action bidding system for adsML designAdvertisers pay for actions (installs, purchases), not clicks, and set a target cost per action. Design the system that bids in each ad auction to hit that target: predicting…
Last asked Mar 2026
Analyze the failures of a vision modelML designA vision model works well on average but fails on some inputs. Explain how you would find and group the failures (slicing by condition, looking at examples, confusion analysis)…
Last asked Mar 2026
Write and speed up an image filter for on-device useML designWrite an image filter (a convolution over pixels, such as a blur) and make it fast enough to run on a device. Discuss the cost per pixel, separable kernels, edge handling, memory…
Last asked Mar 2026
Design a short-video recommendation systemML designDesign the recommendation system behind a short-video feed, where each user swipes through videos one at a time. Cover what you optimise (watch time, completion, likes, shares)…
Last asked Feb 2026
Design an ML platform portalML designDesign the portal that machine learning engineers use to work with the company's ML platform: finding datasets and features, launching training jobs, tracking experiments…
Last asked Feb 2026
QKV attention programming with einsum fluencyML designA short programming screen around sequence models and scaled dot-product attention, plus quick ML knowledge questions. Reading equations and expressing the math with einsum-style…
Last asked Feb 2026
Audio preprocessing and model training choicesML designCompare ways to preprocess audio for a model (raw waveform, spectrograms, mel features, normalization, augmentation) and their pros and cons, then walk through the steps of…
Last asked Feb 2026
Self-contained ML take-home with live reviewML designA few-hour modeling problem you solve alone on your own machine, followed by a live review: you present your results, code and trade-offs, and the interviewers ask questions…
Last asked Feb 2026
Modeling other agents in a driving simulatorML designIn a simulator used to test a self-driving stack, how do you model the other agents (cars, pedestrians, cyclists)? Compare replaying logged behavior, rule-based agents and learned…
Design an image copyright-violation detection systemML designDesign a system that detects when users upload images that infringe someone's copyright on a large social platform. Discuss where the reference set of protected images comes from…
Last asked Feb 2026
LLM fine-tuning and generative modelsML designExplain how large language models are fine-tuned: supervised fine-tuning, parameter-efficient methods such as LoRA, and preference tuning, with the trade-offs of each. Then…
Clean OCR output into a training dataset for an LLMML designYou have a large amount of text produced by OCR. Design the pipeline that cleans it and turns it into a dataset for training a language model: fixing or dropping garbled text…
Last asked Feb 2026
Design a video vision-language model end to endML designDesign a model that understands video and answers questions about it in language: data and its labels, frame sampling and the video encoder, how visual tokens reach the language…
Last asked Feb 2026
Design app recommendations for an app storeML designRecommend apps to each user on an app store's home page and in its search results. Cover candidate generation and ranking, the signals you would use (installs, uninstalls, usage…
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
Design an online experimentML designDesign an online experiment (A/B test) for a product or model change. Cover the hypothesis and metrics, randomization unit, sample size and duration, guardrail metrics, how you…
Last asked Feb 2026
Prompt design with evaluation scaffoldingML designIn a notebook, design prompts for a defined task and write the surrounding scaffolding that evaluates them. The exercise tests task framing and measurement more than clever…
Last asked Jan 2026
Design a model inference pipelineML designBilled as ML design but really ML infrastructure: design a pipeline that serves a trained model's predictions. Cover how requests or data arrive, batching, GPU use, model versions…
Last asked Jan 2026
Design a notification ranking systemML designMany products send notifications to the same user: new products, friends' activity, results of advertisers' campaigns and more. Design the ML system that decides which…
Last asked Jan 2026
Design an ads ranking system with calibrated predictionsML designDesign the system that ranks ads for a user, using predicted click and conversion probabilities together with advertiser bids. Because the predictions feed an auction, they must…
Last asked Jan 2026
Shard a large input embedding table across devicesML designA model's V, H input embedding table is too big for one device. Compare sharding along the vocabulary with sharding along the hidden dimension: how a lookup works in each, the…
Last asked Jan 2026
Shard an LM head and combine the logitsML designThe output layer maps a hidden state through an H, V weight matrix to vocabulary logits. Compare splitting the weight by vocabulary with splitting it by hidden dimension: what…
Last asked Jan 2026
MLP tensor shapes and PyTorch broadcastingML designGiven a small multi-layer perceptron and a batch of inputs, write down the shape of every weight, bias, activation and gradient. Then explain PyTorch's broadcasting rules, how the…
Last asked Jan 2026
Infer a user's intent from their typing in real timeML designAs a user types into an assistant or search box, predict what they intend to do so the product can respond early. Cover labels and data, features from partial input, the model and…
Last asked Jan 2026
Debug a model that does worse online than offlineML designA model scored well in offline evaluation, but after launch its online metrics are worse. How do you find out why? Walk through the likely causes (training-serving skew in…
Last asked Jan 2026
Design photo and listing quality modelsML designDesign models that score the quality of listing photos and of listings overall, so better listings can be surfaced and hosts can be told what to improve. Cover defining quality…
