What is Interview Forge?

Interview Forge helps you practice mock interviews, follow study plans, track career goals, review learning analytics, and prepare across 36 pages.

Design Twitter (Timeline & Feed) at Amazon — System Design Interview Guide

The problem at Amazon

Functional: post a tweet, follow users, see a home timeline, see a user timeline, search.
Non-functional: timeline loads in <100ms, posts appear in seconds, the system handles celebrity fan-out.
Scale: ~300M users, ~500M tweets/day, read-heavy (timeline reads >> writes).
Consistency: timelines can be eventually consistent (a tweet may take a second to appear); ordering must be monotonic.

Out of scope (covered in detail below)

Trending topics, ads, live video, Spaces (audio), and lists — deferred and discussed in the dedicated section.

At Amazon, this problem is framed around customer-obsessed, data-driven, bar-raising. Expect Amazon-specific scale numbers and a focus on production readiness and the company's engineering bar.

Capacity estimation

Resource	Estimate	Notes
Tweets/day	~500M	Immutable, small
Timeline reads/day	~billions	Read-heavy, ~100x writes
Timeline cache (Redis)	~terabytes	Pre-built home timelines
Max followers	~tens of millions	The celebrity problem

Architecture overview

Tweet write + timeline read

+--------+  tweet   +--------------+   fan-out   +-------------------+
| Client |--------->| Tweet Service|------------>| Timeline Cache    |
+--------+          +--------------+             | (Redis sorted set)|
                          |                     +-------------------+
                          v                              ^
                  +---------------+  follow               |
                  | Graph Service |----------------------+
                  +---------------+
                          |
                          v
                  +---------------+
                  | Search Index  |  (near real-time)
                  +---------------+

Writes fan out to pre-built timelines; reads are a cached list fetch.

Timeline generation

The core decision, identical in spirit to the news feed: where to compute the timeline.

Fan-out on write: push new tweets into followers' pre-built timelines (Redis sorted sets). Fast reads, but a celebrity tweet fans out to millions — write amplification.
Fan-out on read: merge recent tweets from followed users at read time. Cheap writes, heavier reads.
Hybrid: fan-out for normal users, pull-on-read for celebrities (millions of followers) to avoid write amplification — the production answer.

Unlock the full guide

Premium members get every DSA, system design, company and role interview guide — plus AI practice, mock interviews, and analytics.

Unlock with Premium