HLD · CS Fundamentals
HLD (High-Level Design) Interview Guide
HLD interviews ask you to design a system at the architecture level — sharding a search index, handling a network partition, designing failover for a single-leader node — trading off consistency, availability, and latency under real constraints (CAP theorem, not as trivia, but as a forced tradeoff you have to make explicit). The bar is reasoning about failure modes, not naming buzzwords.
What to actually study
- CAP theorem as a forced tradeoff — what you actually give up during a network partition, not the acronym
- Sharding and partitioning strategies for large datasets
- Replication and failover — active-passive, leader election, split-brain
- Probabilistic data structures at scale — Bloom filters, and what they can't guarantee
Real questions from Ediky's bank
From the shared System Design topic, these lean HLD (distributed systems, not single-class design):
- "Design Active-Passive Failover for a Single-Leader Order-Matching Node"
- "A Trade Ledger During a Network Partition — What Does CAP Force You to Drop?"
- "Design a Fast 'Is This Username Taken?' Check for 500M Users"
- "What Can and Can't a Standard Bloom Filter Do?"
How to study on Ediky
Start with Ediky's 26-lecture HLD video course, then use the distributed-systems-flavored questions in CS Fundamentals' System Design topic for applied practice.
Start the HLD video course
26 lectures covering sharding, replication, and CAP-theorem tradeoffs for interviews.
Open Learning Resources