Filter Contacts by Email Domain

You are given contact rows and a requested email domain. Each contact row has the form name,email,phone.
Return every original contact row whose email has exactly one @, a non-empty local part, and a domain equal to the requested domain using case-insensitive comparison. The requested domain may optionally begin with @.
Preserve the contacts' input order. Ignore malformed contact rows and malformed email addresses.

Function
filterContactsByDomain(contacts: String[], domain: String) → String[]

Examples
Example 1
contacts = ["Alice,alice@example.com,111","Bob,bob@other.com,222","Cara,cara@EXAMPLE.COM,333"]
domain = "example.com"
return = ["Alice,alice@example.com,111","Cara,cara@EXAMPLE.COM,333"]
Alice and Cara have the requested domain. Domain comparison is case-insensitive, and their original rows and order are preserved.
Example 2
contacts = ["A,a@shop.io,1","B,b@sub.shop.io,2","C,invalid,3"]
domain = "@shop.io"
return = ["A,a@shop.io,1"]
The optional leading @ is ignored. A subdomain is not an exact match, and the malformed email is ignored.
Example 3
contacts = ["A,@example.com,1","B,b@@example.com,2","C,c@example.org,3"]
domain = "example.com"
return = []
The first email has an empty local part, the second has two @ characters, and the third uses another domain.

Constraints
0 <= contacts.length <= 100000.
Every contact string has length at most 500.
A well-formed contact row contains exactly two commas; its fields do not contain commas.
domain is non-empty after removing one optional leading @ and contains no whitespace, comma, or @.
Email local parts and domains are compared without trimming; only domain letter case is ignored.

The pattern is plain string parsing plus a single pass filter. Normalize the requested domain first: strip one leading @, then lowercase it. For each row, split on commas and require exactly three fields. Then check the email: exactly one @, a non-empty local part, and a lowercased domain that equals the target. Use equality, never endswith, or sub.shop.io sneaks through. Return the original row string, not a rebuilt one, and keep input order by appending as you go. The pitfalls are counting @ characters wrong, forgetting the empty local part like @example.com, and lowercasing the whole email instead of just the domain. Don't trim anything. Runs in O(n * 500), which is fine for 100000 rows. If you freeze under the clock, StealthCoder can hand you the clean version during the live OA, but write the checks in this order and you won't need it.

The honest play: practice the pattern, and have StealthCoder ready for the one you didn't see coming.