# Common Mistakes — High-Level & Low-Level Design Interview Problems

Source: https://www.skillbyai.com/en/design-interviews/round-mistakes

> What loses points in design rounds.

## Patterns that hurt candidates

Frequent mistakes: jumping into a diagram or code before agreeing on scope; designing for imaginary scale (sharding a system that fits on one machine); naming technologies instead of explaining **why** ("use Kafka" with no reason); ignoring the interviewer's hints; spending the whole round on one component; in LLD, creating a single god class, overusing inheritance where composition fits, or forcing design patterns in for show; and never mentioning failure, concurrency or edge cases. Silence is also costly: interviewers can only credit reasoning they hear.

## Weak versus strong phrasing

How to justify a choice.

```text
weak:   "We will use Cassandra."
strong: "Writes dominate (assumed ~10x reads), access is by user id,
         and we can accept eventual consistency for this feed, so a
         wide-column store partitioned by user id fits. If we needed
         multi-row transactions I would choose a relational store."

weak:   class ParkingLot { ...40 methods... }
strong: ParkingLot -> Floors -> Spots; pricing in a separate
        PricingStrategy; ticketing in TicketService
```

## Check in at milestones

After requirements and after the first diagram, ask whether the interviewer wants you to go deeper anywhere. It keeps you aligned with what they score.

**Quiz:** Which is a common LLD mistake?

- [ ] Mentioning thread safety
- [ ] Separating pricing into its own strategy
- [ ] Asking about requirements
- [x] Putting most logic in one large god class

*Answer:* Putting most logic in one large god class. Responsibilities should be split.