← Openai Interview Insights

Openai·Machine Learning Engineer·Onsite - System Design / Architecture·Senior

SeniorPrefer not to say
Jun 2026

Summary

Interviewed for an ML Engineer role at OpenAI and got hit with a massive system design question covering basically the entire lifecycle of an enterprise document search product. Single question but it sprawled into a dozen sub-topics. Left feeling like I'd answered maybe 60% of what they were actually probing for.

Questions Asked (1)

Q1

Design an ML-powered enterprise document search system. Cover requirements like latency SLA, data freshness, multi-tenant access control, and PII handling. Walk through the indexing pipeline, feature generation, storage, hybrid retrieval, ranking, query understanding, personalization, evaluation, monitoring, and how you enforce ACLs at query time without leaking documents.

System DesignTechnical Trade-offsA/B Testing & Experimentation
Author's notes

This question just kept going.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements and constraints (latency SLA, data freshness, multi-tenancy, PII) to frame the design. Then walk through the end-to-end system architecture, emphasizing the indexing pipeline, hybrid retrieval, ranking, and ACL enforcement. Finally, discuss evaluation, monitoring, and trade-offs, showing how each component meets the requirements.

Pro tip: Emphasize that ACLs must be enforced at query time with document-level filters applied before scoring to prevent leakage, and discuss how to handle PII with encryption and tokenization. Also, highlight the importance of offline/online evaluation and A/B testing for continuous improvement.

1. Clarify Requirements and Constraints

Ask questions to understand latency SLA (e.g., p95 < 200ms), data freshness (e.g., near real-time indexing), multi-tenant isolation, PII handling policies, and scale (documents, queries per second).

2. Design Indexing Pipeline and Storage

Outline the ingestion pipeline: document parsing, PII detection/redaction, chunking, embedding generation, and indexing into a hybrid store (e.g., vector DB + inverted index). Ensure multi-tenant separation via namespaces or metadata filtering.

3. Detail Retrieval and Ranking

Explain hybrid retrieval combining lexical (BM25) and semantic (dense vectors) search, followed by a learning-to-rank model. Include query understanding (intent classification, entity recognition) and personalization (user history, role-based boosts).

4. Enforce ACLs and PII Protection

Describe how to enforce access control at query time: apply tenant and user ACL filters as pre-filters in the retrieval stage, ensuring no unauthorized documents are scored or returned. For PII, use encryption, tokenization, and redaction during indexing and retrieval.

5. Evaluate, Monitor, and Iterate

Define offline metrics (NDCG, MRR) and online metrics (CTR, dwell time). Set up monitoring for latency, freshness, and leakage. Use A/B testing to validate ranking and personalization changes.

Key Points to Mention

  • Hybrid retrieval combining sparse (BM25) and dense (embeddings) methods for robust search.
  • Query-time ACL enforcement with pre-filtering to prevent document leakage.
  • PII handling: detection, redaction, encryption, and tokenization in indexing and retrieval.
  • Multi-tenant isolation via namespaces, metadata filtering, or separate indices.
  • Latency SLA optimization: caching, approximate nearest neighbor (ANN) search, and query pruning.
  • Evaluation framework: offline metrics (NDCG, MRR) and online A/B testing for ranking and personalization.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.