Perplexity

English · Русский · AI Interview Atlas › Companies

AI-native products · International · Loop reviewed 2026-09-26

A secondary guide describes product interviews; a compilation attributes engineering topics to Perplexity. These are preparation leads, not confirmation of a current company-wide loop.

On this page: Interview loop · Questions (15)

Roles covered

• AI Engineering: AI / LLM engineer
• AI Leadership: AI product manager (senior to group)

Interview loop

Stage	What happens	Basis
Product manager path	A 30-minute recruiter screen, a 45-minute product sense case, then five 45-minute rounds: product thinking, analytics and strategy, engineering behavioral, engineering execution and design thinking.	† prep guide or compilation Perplexity AI Product Manager (PM) Interview Guide, Aced (formerly Exponent), retrieved 2026-09-26
Engineering questions	A compilation attributes to Perplexity questions on index freshness, attributing claims to sources, LLM gateways, online evaluation and latency metrics.	† prep guide or compilation AI Engineering Interview Questions Company Wise, Outcome School, published 2026-09-19, retrieved 2026-09-26

Coding: unknown. No company or first-hand source here establishes a coding requirement. (AI Engineering Interview Questions Company Wise, Outcome School, published 2026-09-19, retrieved 2026-09-26, Perplexity AI Product Manager (PM) Interview Guide, Aced (formerly Exponent), retrieved 2026-09-26)

Questions reported for Perplexity

LLM fundamentals

• Compare greedy decoding, beam search and temperature, top-k and nucleus sampling; give a failure mode for each.
  • Knowledge · Asked at: Google and Google DeepMind †, Perplexity †
  • Tests: Whether probability manipulation is connected to output diversity and quality.
  • A strong answer covers:
    • Describe each as a transform: argmax, approximate sequence-likelihood search, logit division by T, top-k truncation, cumulative-mass truncation.
    • Name failures: greedy and beam repeat or go bland, high temperature derails, top-k ignores distribution shape, top-p admits flat tails.
    • Choose by task, deterministic for extraction and sampled for generation, and measure accuracy together with diversity across seeds.
  • Read: The Curious Case of Neural Text Degeneration (Holtzman et al., arXiv)

Inference, serving and cost

• Distinguish time to first token, time per output token, inter-token latency and throughput when comparing serving systems.
  • Knowledge · Asked at: Microsoft †, Perplexity † · ✍ Answer
  • Tests: Whether user-visible delay is separated from aggregate capacity.
  • A strong answer covers:
    • Define TTFT as queueing plus prefill, TPOT and inter-token latency as decode pace, throughput as aggregate capacity.
    • Explain how larger batches trade each user's decode speed for throughput, and how admitted prefills stall decodes.
    • Compare at matched load and length distributions: p50 and p99 per latency metric, throughput within the targets.
  • Read: vLLM documentation (vLLM project)
• After a deployment, p99 latency doubles although model weights are unchanged. How do you isolate the cause?
  • Applied scenario · Asked at: Amazon †, Databricks †, OpenAI †, Perplexity † · ✍ Answer
  • Tests: Whether diagnosis separates queueing, scheduling, dependencies and traffic changes.
  • A strong answer covers:
    • Compare traffic mix, queue time and per-stage latency before and after release.
    • Inspect cache hit rates, batching, dependency tails and resource saturation.
    • Reproduce one hypothesis at a time and use a rollback or canary to verify causality.
  • Read: The Tail at Scale (Dean and Barroso, Google Research)

RAG and retrieval

• When would you combine lexical and dense retrieval, and how would you merge their rankings?
  • Knowledge · Asked at: Microsoft †, Perplexity † · ✍ Answer
  • Tests: Whether exact terms and semantic similarity receive appropriate treatment.
  • A strong answer covers:
    • Explain lexical strength on identifiers and dense strength on paraphrases.
    • Fuse rankings without assuming raw scores are comparable.
    • Ablate each retriever and reranking on a fixed relevance set.
  • Read: Contextual Retrieval in AI Systems (Anthropic)
• Where should a cross-encoder reranker sit in a retrieval pipeline, and when does its quality gain justify latency?
  • Knowledge · Asked at: Microsoft †, Perplexity † · ✍ Answer
  • Tests: Whether candidate count, recall and reranking cost are balanced.
  • A strong answer covers:
    • Retrieve a broad candidate set first, then score query-document pairs with the reranker.
    • Measure initial recall separately: a reranker cannot recover a document missing from its input.
    • Sweep candidate and final-context counts against held-out answer quality and end-to-end latency.
  • Read: Contextual Retrieval in AI Systems (Anthropic)
• Design continuous index updates, including changed documents, deletions and queries during a rebuild.
  • System design · Asked at: Perplexity † · ✍ Answer
  • Tests: Whether freshness has consistency guarantees and an observable lag.
  • A strong answer covers:
    • Apply ordered changes idempotently by document ID and version; define when a complete new chunk set becomes visible.
    • Enforce deletions and revoked access at query time; rebuild from a consistent snapshot, catch up changes and validate before switching the read alias.
    • Measure source-to-search lag and reconcile counts, representative queries and permissions; rollback must not resurrect deleted or forbidden content.
  • Read: Aliases (Elastic) · Document-level access control (Microsoft Learn)
• How would you connect each factual claim in an answer to the retrieved passage that supports it?
  • System design · Asked at: Perplexity † · ✍ Answer
  • Tests: Whether a citation is checked for support rather than merely attached.
  • A strong answer covers:
    • Bind each claim to a passage identifier during generation instead of appending links at the end.
    • Verify that the cited passage entails the claim before the answer is shown.
    • Define behaviour when nothing supports a claim: drop it, mark it unsupported or abstain.
  • Read: Patterns for Building LLM-based Systems & Products (Eugene Yan)

Evaluation and observability

• Design an LLM judge and a calibration process that exposes order, verbosity and self-preference biases.
  • System design · Asked at: Perplexity † · ✍ Answer
  • Tests: Whether model judgements are validated against independent labels.
  • A strong answer covers:
    • Define a task-specific rubric and a held-out human-labelled calibration set.
    • Randomise answer order and test sensitivity to length and model identity.
    • Measure disagreements and use expert review where automated judgement is unreliable.
  • Read: Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (Zheng et al., arXiv)
• Design online evaluation for an AI feature: choose what to log, review manually and compare in an experiment.
  • System design · Asked at: Perplexity †
  • Tests: Whether experimental outcomes are distinguished from passive correlations.
  • A strong answer covers:
    • Log inputs, outputs, versions and downstream outcomes per request, within consent and retention limits.
    • Review a stratified sample manually to find failure types, but treat logged correlations as hypotheses.
    • Establish causal effects with a randomised experiment, a pre-registered primary metric, guardrails and adequate power.
  • Read: Your AI Product Needs Evals (Hamel Husain) · Measuring the User Experience on a Large Scale: User-Centered Metrics for Web Applications (Rodden, Hutchinson and Fu, Google Research)

AI system design

• Design an LLM gateway with provider routing, failover, caching, rate limits and enforceable spending budgets.
  • System design · Asked at: Palantir †, Perplexity † · ✍ Answer
  • Tests: Whether provider differences and failure handling preserve the service contract.
  • A strong answer covers:
    • Define provider capabilities and normalise only compatible request semantics.
    • Enforce tenant quotas and budgets before dispatch, including retries.
    • Specify failover boundaries for partially streamed responses and side-effecting tool calls.
  • Read: Scaling your API with rate limiters (Stripe) · Designing robust and predictable APIs with idempotency (Stripe)

Practical coding

• Write an asynchronous API batch processor with bounded concurrency, retry jitter and isolated per-item failures.
  • Coding · Asked at: Anthropic †, Perplexity † · ✍ Answer
  • Tests: Whether cancellation, retries and results remain well-defined under partial failure.
  • A strong answer covers:
    • Bound in-flight work and propagate cancellation and deadlines.
    • Retry only suitable failures with capped backoff and jitter; consider idempotency.
    • Return per-item outcomes and test timeouts, permanent failures and partial completion.
  • Read: Designing robust and predictable APIs with idempotency (Stripe) · Site Reliability Engineering (Google)

AI product strategy and metrics

• What North Star metric would you choose for an AI search product, and what would constrain it?
  • Product case · Senior · Asked at: Perplexity † · ✍ Answer
  • Tests: Whether one headline metric remains accountable to trust and task completion.
  • A strong answer covers:
    • Define a valuable completed search task and how to observe it.
    • Add quality, safety, latency and cost guardrails.
    • Explain denominator, segmentation and how the metric could mislead.
  • Read: Measuring the User Experience on a Large Scale: User-Centered Metrics for Web Applications (Rodden, Hutchinson and Fu, Google Research)
• Propose a new feature for Perplexity and explain whom it serves.
  • Product case · Senior · Asked at: Perplexity †
  • Tests: Whether you can make a differentiated, evidence-seeking product proposal.
  • A strong answer covers:
    • Pick one segment and a recurring research job that users still finish outside the product.
    • Explain why rivals cannot simply copy it, and measure completed tasks with a citation-accuracy guardrail.
    • Test the riskiest assumption cheaply with a fake-door test or a manual prototype for target users.
  • Read: Measuring the User Experience on a Large Scale: User-Centered Metrics for Web Applications (Rodden, Hutchinson and Fu, Google Research)
• Where does an AI answer engine outperform traditional search, and where does it fall short? What would you improve first?
  • Product case · Senior · Asked at: Perplexity † · ✍ Answer
  • Tests: Whether you compare products through user tasks and evidence quality.
  • A strong answer covers:
    • Split queries by task: navigation, a single fact, comparison across sources, open exploration.
    • Judge an answer by whether its evidence is checkable, current and correctly attributed.
    • Name the first improvement and the metric that would show it worked.
  • Read: Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (Lewis et al., arXiv)

Leading engineering teams

• How would you convince engineers that a proposed product feature is worth building?
  • Applied scenario · Senior · Asked at: Perplexity †
  • Tests: Whether you engage technical concerns and establish a shared reason to invest.
  • A strong answer covers:
    • Bring the user problem and the evidence behind it, not the feature, so engineers can judge the reasoning.
    • Treat objections about cost, complexity or maintenance as inputs that can reshape scope, not as resistance.
    • Agree a success metric and a cheap first version, and commit to cutting it if the metric misses.
  • Read: User Needs + Defining Success (Google PAIR) · Design Docs at Google (Industrial Empathy)

← Cursor (Anysphere) · Sierra →