"If you have 1 hour a day, this is how you crack L5/L6 at Google/Meta/Amazon."
This roadmap is designed by a Senior Staff Engineer to take you from "knowing the terms" to "designing exabyte-scale systems."
Focus: Mastering the 'Why'. Do not skip this. Most L5 rejections happen here.
- Topics: Vertical vs Horizontal scaling, Load Balancing (L4 vs L7), Consistent Hashing.
- Goal: Explain why round-robin LB is a bad idea for stateful services.
- Checklist: Read
02-fundamentals/02-scalability-basics.md.
- Topics: B-Trees vs LSM-Trees, SQL vs NoSQL (The real tradeoff), Indexing strategies.
- Goal: Choose between Postgres and Cassandra based on write-amplification.
- Checklist: Read
03-building-blocks/02-databases-sql-nosql.md.
- Topics: CAP Theorem (The PIE version), PACELC, Consensus (Raft/Paxos).
- Goal: Explain how Spanner achieves "Strong Consistency" globally.
- Checklist: Read
02-fundamentals/04-consistency-patterns.md.
Focus: Designing the interface and the data flow.
- Topics: Cache-aside vs Write-through, Redis architecture, Kafka partitioning.
- Goal: Design a system that handles 1M QPS using only 10% of the DB capacity.
- Checklist: Read
03-building-blocks/03-caches.md.
- Topics: REST vs gRPC, Protobuf, Schema design for NoSQL.
- Goal: Design an API that is backward-compatible for 3 years.
- Checklist: Create 3 schemas for your favorite app.
- Topics: HTTP/2 vs HTTP/3, DNS (Anycast/Latency-based), OAuth2 flows.
- Goal: Explain how a CDN resolves "Cold Start" latency for global users.
Focus: Putting it all together. 2 Case studies per week.
- Projects: Design a URL Shortener, Design a Notification System.
- Focus: Efficiency and durability.
- Projects: Design Twitter/X, Design YouTube.
- Focus: Fan-out, Storage optimization, and Ingestion pipelines.
- Projects: Design Uber (Real-time GIS), Design a Payment System (Distributed Transactions).
- Focus: High availability and Data integrity.
Focus: Failure, Post-mortems, and Delivery.
- Topics: Multi-region Active-Active, Chaos Engineering, Load Shedding.
- Goal: Design a system that can lose an entire AWS Region and heal in < 60s.
- Topics: Cache Stampede, Thundering Herd, Cascading Failures.
- Goal: Walk through a Meta outage and explain the "Fix" on a whiteboard.
- Goal: Do 3 mocks on Pramp or with a colleague.
- Checklist: Use the
PRACTICE-CHECKLIST.mdto grade yourself.
Design a system that ranks 1B ads per second across 3 continents with <50ms P99 latency. If you can answer this without sweating, you are ready for Staff (L6).
For those who need faster preparation (interview in 6 weeks)
| Day | Topic | Practice |
|---|---|---|
| 1-2 | CAP Theorem, PACELC | Explain with real examples |
| 3-4 | Database: SQL vs NoSQL | Choose DB for 3 scenarios |
| 5-6 | Caching strategies | Design cache for Twitter |
| 7 | Message Queues | Design notification system |
| Day | Topic | Practice |
|---|---|---|
| 8-9 | API Design | Design CRUD API for e-commerce |
| 10-11 | Data Modeling | Design schema for Uber |
| 12-13 | Load Balancing, CDNs | Design CDN architecture |
| 14 | Mock Interview 1 | Record and review |
| Day | Topic | Practice |
|---|---|---|
| 15-17 | Twitter/X Design | Full design with 2 mocks |
| 18-20 | Payment System | Focus on consistency |
| 21-23 | Multi-region, Chaos | Design DR strategy |
| 24 | Mock Interview 2 | Feedback from peer |
| 25-27 | Real-world failures | Study 5+ postmortems |
| 28 | Mock Interview 3 | Timed practice |
- Focus on completing happy path
- Master 4 easy case studies
- Practice speaking while drawing
- Master all 16 case studies
- Quantify every decision
- Handle 3+ follow-ups
- Multi-region design
- Cost modeling
- Organizational thinking