When you smell this pattern
- Need pairs/complements → hash map of seen values
- Counts / anagrams → frequency map
- Subarray sums → prefix sums + map of prefixes
- “First unique” → count then scan
Template (complement)
```
map = {}
for i, x in enumerate(nums):
need = target - x
if need in map: return [map[need], i]
map[x] = i
```
Interview tip
State why a nested loop fails the constraint **before** writing the map.
Practice
Two Sum + Top K Frequent in the bank. Narrate the hash tradeoff.