Microsoft Software Quality Engineer Intern Interview:
Process + Questions

Pick 2 to 3 things relevant to quality or engineering. Don't read your resume back to them. Tell a short story about each and what you took away.

Be specific. Which products do you use? Which team interests you? Generic answers get forgotten. Talk about something real.

Show that you listen, understand the other side, and find resolution. They want to hear you made the team work better, not that you won the argument.

Pick a real technical or process challenge. Walk through what went wrong, what you tried, what you learned. The learning matters more than the fix.

Doesn't need to be a formal leadership role. Starting a study group, proposing a process change, fixing something nobody asked you to fix all count.

Pick something you actually use. Be specific: 'I would add X because Y' is better than vague feature requests. Shows product awareness.

Two pointers from each end, swap inward. Or slicing in Python. O(n) time, O(1) space for in-place.

Hash map for frequencies, iterate to find the max. O(n). Mention how you'd handle ties.

Hash map storing complements as you iterate. One pass, O(n) time, O(n) space. Mention edge cases: duplicates, no solution.

Floyd's algorithm. Slow pointer moves 1 step, fast moves 2. If they meet, cycle exists. O(n) time, O(1) space.

Recursive DFS. Depth is 1 + max(left, right). Base case: null returns 0. Could also do BFS level by level.

BFS for unweighted grids. Dijkstra if edges have weights. Time: O(rows × cols) for BFS.

In-order traversal should give strictly increasing values. Or recursive min/max bounds on each node.

Two stacks: main stack for values, auxiliary tracks the minimum at each level. O(1) for all operations.

Functional: valid creds work, invalid don't. Edge: empty fields, SQL injection, XSS, max-length. Security: brute force lockout, HTTPS, session expiry. Accessibility: screen reader, keyboard nav.

Unit (individual functions), integration (components together), system (full app), regression (after changes), UAT (end users validate). Know when each is used.

Requirements analysis → test planning → test case design → environment setup → execution → defect reporting → retesting → closure. Know what comes out of each phase.

Scope the key features and users. Test: exact match, partial match, no results, special chars, load performance, relevance ranking. Non-functional: response time, accessibility, localization.

Risk-based. Focus on highest user impact and highest defect likelihood. Smoke test everything else. Automate the repetitive stuff so manual time goes to exploratory.

Severity = how bad technically (crash vs typo). Priority = how urgently to fix (business impact). A homepage typo is low severity but could be high priority.

Testing existing features after code changes. Every fix can introduce side effects. This is why automated regression suites exist and keep growing.

Black box: test from inputs/outputs without seeing code. White box: test based on internal logic and code paths. QA mostly does black box. Devs write white box unit tests.