Pinterest·Software Engineer·Onsite - System Design / Architecture·Senior

SeniorPrefer not to say

Jun 2026

Summary

Pinterest system design round, one big question about building a blob storage service from scratch. Pretty intense scope, they wanted end-to-end coverage across a lot of dimensions and weren't shy about pushing on trade-offs.

Questions Asked (1)

Design a large-scale, highly available blob storage service similar to Amazon S3, covering APIs, data model, architecture, partitioning, durability, consistency, large object handling, security, and monitoring.

System DesignTechnical Trade-offsData Modeling

Author's notes

This one sprawls in every direction and I felt it immediately.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements (scale, consistency, durability, access patterns) and then present a high-level architecture before diving into each component. Structure your answer around the key areas mentioned: APIs, data model, partitioning, durability, consistency, large objects, security, and monitoring, making explicit trade-offs at each step.

Pro tip: Emphasize how your design handles failures and scales horizontally, and relate it to Pinterest's specific needs like storing billions of images with low latency and high durability. Mention real-world examples like S3's design or Google's Colossus to show depth.

1. Clarify Requirements and Scope

Ask questions to understand expected scale (e.g., petabytes, billions of objects), read/write patterns, consistency requirements, durability targets (e.g., 11 9's), and security/compliance needs. This ensures your design addresses the right problems.

2. Define APIs and Data Model

Outline core operations: PUT, GET, DELETE, LIST, and possibly multipart upload. Describe the data model: buckets, objects with metadata, and unique identifiers. Consider how to handle large objects via chunking.

3. Design High-Level Architecture

Propose a layered architecture: front-end API servers, metadata service, data storage nodes, and a distributed coordination layer. Discuss how to achieve high availability and durability through replication and erasure coding.

4. Address Partitioning, Consistency, and Durability

Explain partitioning strategies (e.g., consistent hashing) to distribute data and metadata. Discuss consistency models (e.g., eventual vs. strong) and how to achieve durability via replication across availability zones and erasure coding.

5. Cover Security, Monitoring, and Trade-offs

Detail security measures: authentication, authorization, encryption at rest and in transit. Describe monitoring for availability, latency, and durability. Summarize key trade-offs (e.g., consistency vs. latency, cost vs. durability).

Key Points to Mention

Durability techniques: replication (e.g., 3x) and erasure coding (e.g., Reed-Solomon) with trade-offs in storage overhead and repair cost.
Consistency models: strong consistency for metadata operations vs. eventual consistency for data, and how to handle read-after-write consistency.
Partitioning: consistent hashing or range partitioning for metadata and data, with strategies to avoid hotspots and enable scalability.
Large object handling: chunking into fixed-size blocks, multipart uploads, and parallel transfers for performance.
Security: IAM policies, bucket policies, encryption (SSE, TLS), and audit logging.
Monitoring: metrics for request latency, error rates, storage utilization, and durability (e.g., via checksums and background scrubbing).

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