karthik.dev

Engineering Showcase

System Design Gallery

Interactive, pan-and-zoom architecture diagrams covering large-scale system design patterns, plus AI agent and multi-tenant SaaS architectures. Click any service to see its role, and check the panel for traffic estimates, database choices, scaling decisions, and trade-offs.

Note: the well-known-company-inspired diagrams below are generic, publicly-documented conceptual patterns for teaching system design — illustrative approximations, not verbatim internal architectures.

Drag to pan · Scroll buttons to zoom · Click a node for details
video segmentsbrowse/searchpublish renditionsClient AppsCDN Edge CacheAPI GatewayRecommendation Serv…Catalog ServiceEncoding PipelineViewing History Sto…Billing DB

Video Streaming Platform (Netflix-style)

A globally distributed video-on-demand system optimized for low-latency playback start and adaptive bitrate streaming at massive scale.

Click any node in the diagram to inspect it.

Traffic Estimate

Hundreds of millions of daily active viewers; peak concurrent streams in the tens of millions.

Database Choices

  • Cassandra / wide-column store for viewing history and recommendations at write-heavy scale
  • MySQL/Postgres for billing and account data requiring strong consistency
  • Elasticsearch for title and metadata search

Scaling Decisions

  • Pre-transcode content into multiple bitrates/resolutions and cache on CDN edge nodes close to users
  • Use adaptive bitrate streaming (ABR) so the client adjusts quality to available bandwidth
  • Separate control plane (browse, search, account) from the data plane (video delivery) so each scales independently

Trade-offs

  • Eventual consistency for recommendations/history in exchange for write throughput
  • Heavy CDN pre-positioning cost in exchange for low playback start latency globally