# Design Instagram.

Aced (formerly Exponent) publishes a public search API over its indexed pages; the guide is at https://www.aced.io/agents/agent-access.

Source: https://www.aced.io/questions/1303/design-instagram
Source observed: 2026-11-04T21:42:56.034Z
Last checked: 2026-11-04T21:42:56.034+00:00
Indexed: 2026-11-05T00:58:41.855+00:00

---

# Design Instagram.

This is an Aced interview question page for practicing the system and product design problem **“Design Instagram.”** The question is tagged **Product Design** and **System Design** and is associated with roles including Software Engineer, Engineering Manager, Product Manager, Machine Learning Engineer, and Technical Program Manager.

## Question details

- **Reported interview count:** 15 interviews, plus one additional indicator shown in the page controls
- **Category:** Product Design
- **Last asked:** 2 months ago
- **Save count shown:** 306
- **Roles listed:** [Software Engineer questions](https://www.aced.io/questions?role=swe), [Engineering Manager questions](https://www.aced.io/questions?role=em), [Product Manager questions](https://www.aced.io/questions?role=pm), [Machine Learning Engineer questions](https://www.aced.io/questions?role=ml-engineer), [Technical Program Manager questions](https://www.aced.io/questions?role=tpm)
- **Companies listed:** [Meta](https://www.aced.io/questions?company=meta-facebook), [Amazon](https://www.aced.io/questions?company=amazon), [DoorDash](https://www.aced.io/questions?company=doordash), [Mark43](https://www.aced.io/questions?company=Mark43), [Instagram](https://www.aced.io/questions?company=instagram)

## Member access

Membership includes:

- 100+ expert answer videos
- Access to all courses
- Peer-to-peer practice
- Private Discord community
- Full question database

[Upgrade now](https://www.aced.io/upgrade?src=questions)

## Interview experiences

The page contains **4 shared interview experiences**:

1. **Amazon · 2 months ago** — Software Engineer · Senior / L5. The interviewer focused on understanding different approaches rather than one particular solution. [Read the experience](https://www.aced.io/experiences/amazon-senior-software-engineer-interview-78e439)
2. **DoorDash · 3 months ago** — Engineering Manager · Staff / L6. The prompt required designing Instagram with photo and video uploads, limits of videos up to **1GB and 1 per day**, photos up to **10MB and 5 per day**, private and public timelines, followed-user posts, and reverse-chronological `created_at` ordering. Scaling issues such as celebrity posts were introduced later. [Read the experience](https://www.aced.io/experiences/doordash-staff-engineering-manager-interview-841483)
3. **Meta · 3 months ago** — Product Manager · Principal / Director / L8+. The candidate described multiple rigorous rounds, on-the-spot problem-solving, and a demanding number of rounds. [Read the experience](https://www.aced.io/experiences/meta-facebook-principal-product-manager-interview-cfd6a4)
4. **Meta · 5 months ago** — Machine Learning Engineer · Senior / L5. The candidate described traditional MLE interviews with AI-enabled coding rounds where Claude or Codex could be used to write code. [Read the experience](https://www.aced.io/experiences/meta-facebook-senior-machine-learning-engineer-interview-54c669)

## Community answers

There are **39 community answers**. The accessible answers discuss how to frame the problem, product scope, feeds, media delivery, social graphs, search, scale, and trade-offs.

### Thomas F.: clarify what “design Instagram” means

Thomas F. begins by asking what designing Instagram actually means. [Thomas F.](https://www.aced.io/users/95228779-9dcb-42b3-994f-418cba9ebc38) has 171 karma and posted March 10, 2025.

The response recommends recognizing that Instagram encompasses multiple products and surfaces, including Stories, Reels, Threads, and messaging, rather than treating it only as a collection of posts. It also identifies these areas for discussion:

- **Business:** Instagram is an advertising business, so fast, well-targeted ad serving and the signals used for targeting are important design considerations.
- **Competition and ranking:** Basic post display is easy to copy; algorithms that select and rank content are a key area for differentiation.
- **Connected versus disconnected content:** The design can consider friend content, recommended content, and messaging interactions, with implications for both user experience and system design.
- **Creators:** Creators may need tools to create posts and Reels, schedule content, and collaborate.

The response also suggests these sub-problems:

- Fast feed delivery, including prefetching and reordering based on current interaction signals.
- A global CDN for large volumes of media.
- Modeling the social graph and extracting signals from relationships.
- The image lifecycle: client-side preprocessing, server-side processing, signal extraction, content review queues, CDN tiers, and regional cache placement.

It argues that a one-hour high-level design should address broader systems and pipelines rather than concentrating only on post-table columns.

### Scott S.: a distributed architecture proposal

Scott S. posted an answer updated July 20, 2024. [Scott S.](https://www.aced.io/users/dcbeea3a-839a-4bfb-a6e2-66a66b916b5b) has 1,034 karma.

The response proposes a distributed web-server and database architecture with a load balancer. It suggests partially loading each user’s newsfeed with recent posts first, plus dedicated servers and caching for users with very high follower counts.

#### Functional requirements listed

- Photo upload
- Viewing by user link for sharing
- Delete, tag, modify, and captions
- Social actions: comments, sharing, votes, friends, and subscribers
- Tag search, with reverse-chronological results and possible friendship weighting

#### Capacity figures in the response

The response uses a design scenario of **100M DAU**, about **50 images viewed per user per day**, **5B image views per day**, and approximately **50k views per second**. It estimates **100M new photos daily** if each user uploads one per day, approximately **500 TB/day** at 5 MB per photo including thumbnails, and **25 PB/day** of read traffic at a 50× read-to-write ratio. These are figures from the community design scenario.

#### Proposed storage and delivery approach

1. Prefer AP in the CAP trade-off so photos and videos are saved even if every user does not immediately see every new post.
2. Use eventual consistency to synchronize databases.
3. Use a NoSQL database keyed by user account, containing follower keys and photo keys.
4. Use an ID-indexed SQL database for photos, sharded using an immutable account identifier.
5. Add new photos to cache immediately and use asynchronous processing through a message queue.
6. Use a master–slave database design for reliability.
7. Use a load balancer to distribute requests and select the relevant shard.
8. Use CDN caching with eviction heuristics based on nearby follower counts and cache age; fetch missing photos from the database and add them back to the cache.

#### Metadata, comments, tags, and search

- Store metadata such as upvotes and captions in a relational database, with short-lived caching for popular-picture totals.
- Store comments in a normalized relation such as `(imageId, commenterId, comment)` and paginate comment queries.
- Store user information separately in a table or document database.
- Index photos by tags and rank results using time, popularity, and friendship.
- Use a search pipeline of **candidate generation**, **scoring**, and **reranking**. Candidate generation can use tag indexes and friends’ recent posts; scoring assigns relevance; reranking merges candidates into result pages.

#### Scalability and media processing

The response recommends erasure coding and replication for images, leader–follower database replication, cross-data-center replication, CDN and SSD caching, Redis-like caches for metadata/comments/indexes, pre-ranked tag indexes, and many API servers.

It also suggests evicting deleted or modified photos from caches and CDNs, using worker nodes and queues for thumbnails and filters, and verifying that image storage and metadata writes both completed. A “to be verified” queue can track image IDs across concurrent writes. The response identifies storage/retrieval and indexed search as the hardest areas and gives an estimate of **50k–100k modest metadata requests per second** for a large site.

The answer also lists these feature labels:

- **Photo Upload:** upload photos and store them online.
- **Tagging for Search:** add tags for discovery.
- **Basic Filters:** enhance photos with Pillow.
- **Sharing on Social Networks:** share through Facebook or Twitter using special keys.
- **Scalability:** use multiple servers for many users and fast delivery.
- **Addressing File Accessibility Issues:** make copies for simultaneous access to popular photos.

### Other community answers

- **Lindsey P.** — “Great design, thanks!” [Lindsey P.](https://www.aced.io/users/f8622976-3a0e-45be-83ca-5e502bc2d69a) has 7,601 karma and posted March 1, 2022.
- **Anonymous Prawn** — Asked which canvas/editor was used to draw the system-design diagram. [Anonymous Prawn](https://www.aced.io/users/4dc8e03b-63fb-46f1-9834-b8604d068f51) has 165 karma and posted November 15, 2024. The visible reply identifies the tool as **Whimsical**.
- **Justin H.** — Critiqued an approach that spends the first 15 minutes on APIs and data models, recommending that data models, APIs, and capacity calculations arise organically as requirements demand. [Justin H.](https://www.aced.io/users/b3dc0c90-162e-44ca-a044-b04a48cb3bb1) has 332 karma and posted January 5, 2024.
- **Alfred O.** — Replied that the approach addressed core requirements such as uploads and feeds, while suggesting discussion of feed-update intervals, hybrid handling for celebrity posts, APIs, low-bandwidth devices, and image formats.

[Show all 39 answers](https://www.aced.io/questions/1303/design-instagram)

## Related questions

- [Design Instagram's Explore page.](https://www.aced.io/questions/3439/design-instagram-explore-page)
- [Design an Instagram Ranking Model](https://www.aced.io/questions/4012/design-instagram-ranking-model)
- [Design a camera for Instagram users.](https://www.aced.io/questions/4023/design-camera-instagram-users)

## Related courses

### [Machine Learning Engineer Interview Prep](https://www.aced.io/courses/ml-engineer)

![Machine Learning Engineer Interview Prep](https://images.ctfassets.net/x78yjrjc11pq/6UfkHxUyLQVfDHSfLNQtWa/1be1539a65b4f6f15201f5b573c0a578/ML_Engineer_Interview_Prep_Course.jpeg?w=800&h=800&fm=webp&q=75)

Learn from mock interviews, frameworks, and senior-candidate advice. Topics include ML system design, core concepts, coding, behavioral interviews, and more.

### [Engineering Management Interview Prep](https://www.aced.io/courses/engineering-management)

![Engineering Management Interview Prep](https://images.ctfassets.net/x78yjrjc11pq/5AqarwlCDWmr0W7AbMJFwp/16862dab13471f5eb8728b27eb2fbc07/engineering_management_track_compressed_bbabaec4a6__281_29_9a9687c1fb.jpeg?w=800&h=800&fm=webp&q=75)

Prepare for system design, people management, project retrospectives, behavioral interviews, and more.

## Practice

Use the editor to submit your own answer; submitting marks the question complete. You can also [Practice with AI](https://www.aced.io/practice/ai/new?type=PM&question=design-instagram&src=questions).

---

[Read this page’s summary](https://www.aced.io/agents/api/public/v1/page-tldr?url=https%3A%2F%2Fwww.aced.io%2Fquestions%2F1303%2Fdesign-instagram&trace=t_679ff9ab98f3)

## Ask a follow-up about this page

Need a more specific answer? Ask the exact question you have. It returns a fresh, cited answer from Aced (formerly Exponent)’s knowledge base. No account or API key required.

Open this URL with your own question, URL-encoded. Include your question in q. All research inputs are optional; no account or setup.

`GET https://www.aced.io/agents/agent/faq?q={question}&trace=t_679ff9ab98f3`

[Tool guide and examples](https://www.aced.io/agents/agent-access) · [OpenAPI](https://www.aced.io/agents/api/public/v1/openapi.json) · [Page TLDR form](https://www.aced.io/agents/agent-tools/page-tldr?url=https%3A%2F%2Fwww.aced.io%2Fquestions%2F1303%2Fdesign-instagram). Opening a form does not execute a tool.