Step 1 — Invariant: We maintain a window [start, end] that has no repeating characters. We track the last index where each character was seen in a map lastSeen. maxLen stores the maximum window length seen so far.

Step 2 — Expand: For each end from 0 to n - 1, we extend the window to include s[end]. Before doing so, we must ensure the window stays valid (no duplicate). If s[end] was already seen and lastSeen[s[end]] is inside the current window (>= start), then the window would have a duplicate; we move start to lastSeen[s[end]] + 1 so that the duplicate is removed and the window is valid again.

Step 3 — Update: Set lastSeen[s[end]] = end to record that this character was last seen at end. Then update maxLen = max(maxLen, end - start + 1), which is the length of the current valid window.

Step 4 — Result: After processing all indices, maxLen is the length of the longest substring without repeating characters. Each character is processed at most twice (once when end advances, once when start advances), so time is O(n). Space is O(min(n, alphabet size)) for the map.