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

SeniorPrefer not to say

May 2026

Summary

Databricks system design round, one question the whole time: design Slack. Not a lot of other context to share but it was a full session on that alone.

Questions Asked (1)

Design a messaging platform like Slack, covering architecture, real-time communication, and scalability.

System DesignTechnical Trade-offsAPI & Integrations

Author's notes

This is one of those questions where you think you know where to start and then realize halfway through that you've been going too broad.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements and scale, then propose a high-level architecture with clear separation of concerns (e.g., API gateway, message service, presence service, storage). Dive into real-time communication using WebSockets and a pub/sub system, and discuss scalability through partitioning, replication, and caching.

Pro tip: Emphasize trade-offs and justify your choices based on Databricks' data-intensive environment; for example, using Kafka for durable message streaming and Delta Lake for message storage to enable analytics.

1. Clarify Requirements and Scale

Ask about expected user base, message volume, latency requirements, and key features (channels, direct messages, presence, search). Establish assumptions to guide design.

2. High-Level Architecture

Outline core components: API gateway, authentication, message service, presence service, notification service, and storage. Describe how they interact.

3. Real-Time Communication

Explain the use of WebSockets for persistent connections, a pub/sub system (e.g., Kafka) for message fan-out, and how to handle message ordering and delivery guarantees.

4. Scalability and Reliability

Discuss partitioning (e.g., by channel or user), replication for fault tolerance, caching (e.g., Redis for presence), and load balancing. Address horizontal scaling of WebSocket servers.

5. Data Storage and Retrieval

Choose appropriate databases: e.g., Cassandra for messages, Redis for presence, Elasticsearch for search. Discuss indexing, retention, and archival.

Key Points to Mention

WebSocket vs. long polling for real-time communication, and handling reconnections.
Message ordering and exactly-once delivery using Kafka or similar.
Partitioning strategies for channels and users to enable horizontal scaling.
Presence service using heartbeats and Redis with TTL.
Data storage choices: wide-column store for messages, search index for full-text search.
Trade-offs between consistency, availability, and latency (CAP theorem).

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