System Design Syllabus — 24 Weeks

Phase 2. Run in parallel with Neetcode (Phase 1) at 30 min/day, then ramp to 1h/day from Aug.

Resources

• Designing Data-Intensive Applications by Kleppmann (DDIA) — the bible
• System Design Primer (donnemartin/system-design-primer on GitHub)
• HelloInterview free design walkthroughs (YouTube)
• ByteByteGo newsletter + book

Weekly cadence (8 weeks DDIA + 8 weeks SDP + 8 weeks practice)

Block A — DDIA (Weeks 1-8, May–Jun parallel to Neetcode)

Week	DDIA chapter	Topic	Practice
1	Ch 1	Reliable, Scalable, Maintainable	Sketch supplement-store's failure modes
2	Ch 2	Data Models	Document DB vs SQL — why InvoiceAI uses Postgres
3	Ch 3	Storage & Retrieval	B-trees vs LSM trees, indexes (you use Prisma — what indexes does Prisma actually create?)
4	Ch 4	Encoding & Evolution	JSON vs Protobuf vs Avro; how Stripe versions APIs
5	Ch 5	Replication	Single-leader, multi-leader, leaderless. Supabase = single-leader Postgres
6	Ch 6	Partitioning	Why your Life OS uses single-DB partitioned by userId vs sharding
7	Ch 7	Transactions	ACID, isolation levels, Postgres SERIALIZABLE in your billing flows
8	Ch 8 + 9	Distributed problems + Consistency	CAP, consensus (Raft), why this is hard

Block B — System Design Primer (Weeks 9-16, Jul–Aug)

Work through SDP's "Design a system" exercises in order:

Week	Design
9	Pastebin / Bit.ly (URL shortener)
10	Web crawler
11	Mint.com (personal finance — relevant to your Life OS expense tracker!)
12	Twitter / Instagram feed
13	YouTube / Netflix (video streaming)
14	Uber / Lyft (real-time matching)
15	WhatsApp / Slack (chat at scale)
16	Search autocomplete

Block C — Practice + Mock interviews (Weeks 17-24, Sep–Oct)

Week	Activity
17	HelloInterview: Design Dropbox
18	HelloInterview: Design Yelp
19	HelloInterview: Design TopK Ranking
20	First Pramp mock — system design (free, peer)
21	Second Pramp mock
22	First interviewing.io mock with paid Google interviewer ($150-$200)
23	Re-do weakest design from earlier weeks
24	Final mock — record yourself solo on whiteboard, review

The 6-step framework (memorize this)

For any design question:

1. Clarify requirements (5min) — functional + non-functional. Numbers: users, QPS, data size, latency targets.
2. High-level design (5min) — boxes + arrows. Client → API gateway → services → DB.
3. API design (5min) — REST or gRPC, endpoint shapes, request/response.
4. Data model (5min) — tables, relationships, indexes, partition keys.
5. Deep dive (15min) — pick 2-3 components, dive deep. Caching, queue, replication, sharding.
6. Bottlenecks + scaling (10min) — what breaks at 10x? At 100x? How would you fix?

Practice this framework on every design question. Interviewers grade structure as much as substance.

Topics that MUST be fluent

• Load balancing (L4 vs L7, sticky sessions, consistent hashing)
• Caching (Redis vs Memcached, cache-aside vs write-through, TTL strategy)
• Database scaling (read replicas, sharding, denormalization)
• Message queues (Kafka vs RabbitMQ vs Redis Streams; at-most-once vs at-least-once vs exactly-once)
• CDN (when, why, what to cache)
• Rate limiting (token bucket, sliding window log/counter)
• Idempotency (idempotency keys, deduplication)
• Observability (logs, metrics, traces)
• Event sourcing + CQRS basics
• WebSockets / SSE / long polling

Topics you can skip

• Bitcoin/blockchain (irrelevant for L3/L4)
• ML system design (only if applying to ML roles)
• Hardware specifics (you're SWE not SRE)
• Specific cloud quirks (focus on patterns, not "AWS X service does Y")

Building strengthens design (and vice versa)

Every project in your sprint touches a design topic:

• supplement-store: caching strategy for product list, idempotency on Razorpay webhook
• InvoiceAI: queue for reminder cron, multi-tenant data partitioning
• NoteFlow: async processing pipeline, job queue + DLQ for Whisper jobs
• Life OS: modular monolith, observability, BullMQ workers
• Content Engine: distributed cron, fan-out across niches, error handling on third-party APIs

When you read a DDIA chapter, immediately ask "where does this show up in what I'm building?" The answers become interview talking points.