Optimal Card Game Score

A quick note: this problem is backed by a real Spotnana backend interview report. The report directly gave the two-player card game, the choice from either end, optimal play, and Alice's objective, but it did not include an exact function interface, examples, or input limits. The core task match is about 95%.
Alice and Bob play a card game with an array cards. Alice moves first.
On each turn, the current player chooses either the first or the last remaining card and adds its value to their own score. Both players play optimally.
Return the maximum total score Alice can obtain.
Interview Follow-up
The interviewer expected production-ready code written from scratch, including edge-case handling and explicit custom tests.

Function
maxAliceScore(cards: int[]) → int

Examples
Example 1
cards = [1,5,2]
return = 3
If Alice takes 1, Bob takes 5, then Alice gets 2. If Alice takes 2, Bob takes 5, then Alice gets 1. Alice's best total is 3.
Example 2
cards = [1,5,233,7]
return = 234
Alice can force the large middle card into her total and finish with 234.

The trick: define dp[i][j] as the best score difference (current player minus opponent) on cards i..j. Then dp[i][j] = max(cards[i] - dp[i+1][j], cards[j] - dp[i][j-1]). Base case dp[i][i] = cards[i]. Alice's score is (total + dp[0][n-1]) / 2. The pitfall is tracking both players' scores separately, which gets messy. The difference form keeps it clean. Greedy fails on [1,5,2] because taking 2 first hands Bob the 5. Edge cases the interviewer likely probes: empty array returns 0, single card, two cards, negative values if allowed, and odd versus even length. Write your own tests for each. Space can drop to O(n) with a 1D rolling array, but O(n^2) is fine. If the recurrence slips away live, StealthCoder is the hedge that gets you unstuck without anyone seeing it.

If you see this problem in your OA tomorrow, the play is to recognize the pattern in 30 seconds. StealthCoder buys you that recognition.