Google DeepMind AI & ML Engineer interview questions

Optimize any proxy hard enough and the model finds the exploit, not the goal. Sycophancy, padding, and fake citations are reward hacking. Here is why it happens and the controls that actually hold.

CoT is not a universal upgrade. On some tasks it does nothing or hurts, and the reason tells you what to try instead. Here is the escalation ladder past 'let's think step by step'.

For open-ended generation there's no ground-truth string to match, so accuracy is meaningless. The field uses a layered mix of automatic, model-based, and human metrics. Here is how to assemble a credible eval.

Push PPO hard enough and true quality peaks then falls while the reward keeps climbing. The signal is knowing the Gold-vs-proxy gap, the KL budget that bounds it, and how you actually measure when to stop.

GRPO drops the value model and estimates advantage from a group of samples per prompt. The signal is the exact mechanics, why the group baseline works, and the normalization traps that quietly bias training.

Replace a learned reward model with a checker that returns right-or-wrong, and reward hacking mostly disappears. The signal is why verifiable rewards beat learned ones for math and code, and where they break.

The ratio of web, code, books, and math in pretraining quietly decides downstream skills. The signal is knowing how mixtures are chosen, why upsampling helps, and how methods like DoReMi automate it.

If a benchmark leaked into pretraining, the score measures memorization, not ability. The signal is the detection methods (n-gram overlap, canaries, perturbation tests) and why fresh held-out evals matter.

A reward model turns pairwise preferences into a scalar signal RLHF can optimize. The signal is the Bradley-Terry loss, the base-model and head choices, and how you validate it before trusting it.

RLAIF replaces expensive human labels with an LLM's preferences, which scales but inherits the labeler model's biases. The signal is how AI feedback is collected and where it quietly fails.

MQA and GQA shrink the KV cache by sharing key/value heads. MLA takes a different route: compress K and V into a low-rank latent and cache that. The signal is knowing it is a cache trick, not a head-sharing trick, and why it keeps quality.

A model trained at 4K tokens can be stretched to 128K without full retraining. The signal is knowing why naive extrapolation fails and how PI, NTK-aware scaling, and YaRN each rescale RoPE frequencies differently.

Speculative decoding speeds up generation without changing the output distribution. The signal is knowing how draft-model, self-drafting (Medusa/EAGLE), and lookahead approaches differ in where the guesses come from.

FP8 is the format behind modern H100-class training and serving. The signal is knowing the two FP8 variants, why a floating format beats INT8 for dynamic range, and where scaling still matters.

VLMs like GPT-4V and LLaVA bolt a vision encoder onto a language model through a projector. The signal is the image-tokens-as-prefix design, the alignment training, and the resolution and hallucination failure modes.

Speech LLMs either transcribe to text or model audio directly as discrete tokens. The signal is the semantic-vs-acoustic token split and why end-to-end audio models beat ASR-plus-LLM pipelines on latency and prosody.

Everyone budgets image tokens. Almost nobody budgets image latency. Images are a prefill problem, prompt caching stops paying for itself, and the cheapest answer is often not to call a VLM at all.

A VLM that reads an invoice wrong and a VLM that reads it right and reasons wrong produce the same end-to-end score, and they need opposite fixes. Here is how to decompose a multimodal eval, build the set from real traffic, and probe for the hallucinations that accuracy never catches.

Linear agents commit to one path and can't backtrack. Tree-of-Thoughts and LATS add search over reasoning paths. The signal is knowing what they buy, what they cost, and when a cheaper loop wins.

Final-answer accuracy hides how an agent got there. The signal is trajectory-level metrics: tool-selection and argument accuracy, step efficiency, and matching against reference paths, plus when exact-match is the wrong yardstick.

Per-step accuracy looks fine, yet a 50-step task fails. The signal is understanding compounding error and the techniques (decomposition, verification, checkpointing) that keep long-horizon agents from collapsing.

Agents can critique and retry their own work. The signal is knowing how Reflexion's verbal feedback loop works, when self-critique genuinely helps versus when it is theater, and what makes it real.

Agents that drive a browser or screen fail in ways chat agents never do: stale DOM, dynamic pages, wrong clicks. The signal is the grounding and reliability techniques that turn a flaky demo into something usable.

RoPE is why modern LLMs extrapolate to longer contexts, and it's a rotation, not an addition. Coding it shows you actually understand how position enters attention. Here is the implementation.

A from-scratch test of modern optimizers. The signal is the moment estimates, bias correction, and the one change that matters: AdamW decouples weight decay from the gradient instead of folding it into the loss. Here is the implementation and why it wins.

A from-scratch test of how PyTorch actually works under the hood. The signal is building a computation graph during the forward pass, local derivatives per op, and a topological-order backward pass that accumulates gradients. Here is a minimal engine.

The idea that models, like students, learn better from easy examples first is intuitive, and sometimes it works and sometimes it does nothing. Knowing when it pays off is the real signal. Here is the honest answer.

Generative training fails in distinctive ways: GANs collapse to a few outputs, diffusion samples come out noisy or blurry. The symptom tells you which knob to turn. Here is the diagnosis.

Raw gradient saliency maps are noisy and saturate. Integrated gradients fixes both with two axioms and a path integral, but the baseline choice quietly decides the answer. Here is what a careful candidate explains.

An agent that clicks around a real website is slow, brittle, and one wrong click from a costly mistake. The architecture is about perception, action, and guardrails around irreversible steps. Here is how to build it.

'Just set a seed' is the wrong answer. Reproducibility means pinning the environment, the data, the code, and the hardware-level nondeterminism, all at once. Here is the full checklist and where it leaks.

No single parallelism dimension trains a 175B model on its own. The art is layering three of them onto the right hardware so the expensive communication stays on the fast links. Here is how to size each axis.

MoE buys you more parameters for the same FLOPs, but the gating network, the load imbalance, and the all-to-all shuffle introduce failure modes dense models never see. Here is the training playbook.

Data, tensor, and pipeline parallelism all leave one sequence's activations on one device, so 200k+ token training hits a wall none of them can fix. The fourth axis shards the sequence itself, and the interview lives in the communication math.

Large models memorize and can be prompted to emit training examples verbatim, a real privacy and copyright liability. The defenses span data, training, and output. Here is the layered answer.

Gibberish-looking token strings appended to a prompt can reliably break refusals, and they transfer between models. The signal is explaining the gradient search that finds them and why output checks beat input pattern matching.

Every AI team feels the pull between velocity and doing it responsibly. How you navigate that tension, with judgment rather than dogma in either direction, is what this question screens. Here is the answer.

A favorite practical screen at the labs, because it tests concurrency, time handling, and judgment in 20 lines. The trap is the background thread that wastes CPU. Here is the lazy-refill version interviewers want, plus the follow-ups.

A classic that tests whether you reach past sorting for the right structure. The signal is the heap solution (O(n log k)), the bucket-sort O(n) trick, and how it changes for an unbounded stream. Here is the answer.

The canonical grid DP where each cell depends on three neighbors at once. The signal interviewers want is the min-of-three recurrence, and the reason it has to be min. Here is the answer.

A sudden 22-point AUC jump is a gift and a warning. Junior candidates celebrate; strong ones get suspicious and know exactly which leakage checks to run before anything ships.

Moderation is a multi-stage classification problem with brutal tradeoffs: false negatives cause real harm, false positives censor legitimate users, and the adversary adapts. The signal is the tiered pipeline, per-severity precision/recall calibration, and human-in-the-loop. Here is the design.

Autocomplete is the canonical trie question, and it tests whether you reach for the right structure instead of scanning a word list. The signal is O(prefix) lookup, the DFS to collect completions, and the follow-ups (ranking, memory). Here is the build.

Sliding window is one of the highest-yield coding patterns, and this is its canonical problem. The real signal is recognizing when a window collapses an O(n squared) scan into one linear pass, and maintaining the invariant cleanly.

Graph traversal underpins a huge class of interview problems. The signal is knowing the BFS-vs-DFS tradeoff (shortest path vs memory shape) and applying it cleanly, then handling the directed-vs-undirected cycle gotcha most candidates miss.