Design an Image CDN in 2026 — Resize, Cache, and Serve at Scale
Image CDN design: on-demand transformations, format negotiation (AVIF/WebP), cache hierarchy, abuse mitigation, and the architecture that scales.
System Design
Event-Driven Architecture in 2026 — When It's Right and When It's Just Latency
Practical EDA: when async events fit, outbox patterns, idempotency, dead letters, and an honest take on EDA’s costs vs benefits.
System Design Interviews in 2026 — The Working Approach
Practical system design interview: framework, capacity estimates, scaling patterns, and what senior engineers actually look for vs cargo-cult answers.
Designing API Rate Limits Customers Don't Hate — Tiers, Headers, and Fairness
API rate limit UX: tier matrices, X-RateLimit headers, 429 + Retry-After, idempotency interaction, and how Stripe / GitHub do it.
Design a Feed System — Pull, Push, and Hybrid Fanout at Scale
Feed system design: pull (compute on read), push (fanout-on-write), hybrid for celebrities, ranking pipelines, and how feed systems scale.
Design a Chat System — From DM to a Million Users
Chat system design at every scale: message store, conversation index, WebSocket fanout, presence, read receipts, and offline delivery.
Design a Search System — From LIKE Queries to Hybrid Vector Search
Search system at every scale: Postgres FTS for small apps, Elastic / OpenSearch for keyword scale, hybrid (BM25 + vectors) for modern relevance, and ranking discipline.
Design a Payment System — Idempotency, Reconciliation, and the Boring Discipline
Payment system design from first principles: ledger, idempotency keys, reconciliation against PSP, retries, and how to keep money straight at scale.
Design a URL Shortener — From One Box to a Billion Redirects
URL shortener design: base62 keys, ID generation strategies, caching, analytics fanout, and how to handle a viral link gracefully.
Design a Rate Limiter — Token Bucket, Sliding Window, and Distributed Counters
Rate limiter design at scale: algorithms (token bucket vs sliding window), single-node vs distributed, Redis Lua, fair queueing, and operational concerns.