Microsoft software engineer interviews: explain, code and test your solution

Practicing a solution is only part of technical interview preparation. You also need to clarify the problem, explain a choice, write working code and test the cases that could break it. Use the original exercises here to rehearse that complete sequence.

Try three original practice questions

Original Prepti practice, not real or reported employer questions. Optional timing is a practice aid, not an employer test limit. No score prediction or account needed.

Try it: 3 original questions on arrays and boundariesUntimed

Original Prepti practice, not employer questions. Timing is optional; no answer is submitted automatically.

JavaScript is off. Read the original questions and worked answers below; no timer is required.

For a sorted input, what does first_ge([2, 4, 4, 4, 7], 4) return?

def first_ge(a, t):
    lo, hi = 0, len(a)   # hi is EXCLUSIVE
    while lo < hi:
        mid = (lo + hi) // 2
        if a[mid] < t:
            lo = mid + 1
        else:
            hi = mid
    return lo

1: When a[mid] < t the search moves right (lo = mid + 1); otherwise hi contracts to mid. The loop ends with lo on the first index whose value is at least t – here index 1, the first 4. Because hi is exclusive and starts at len(a), the code never reads past the end.

The contract requires a non-empty subarray and non-empty input. What does this function return for a = [-2, -3, -1, -5]?

def max_sub(a):
    best = cur = a[0]
    for x in a[1:]:
        cur = max(x, cur + x)
        best = max(best, cur)
    return best

-1: cur restarts at x whenever x beats extending the run. Every value is negative, so best ends as the largest single element, -1. The classic boundary bug is initializing cur and best to 0, which wrongly returns 0 on all-negative input when a non-empty subarray is required.

A sorted array with distinct values was rotated. What does find_min return for a = [4, 5, 6, 7, 0, 1, 2]?

def find_min(a):
    lo, hi = 0, len(a) - 1   # hi is INCLUSIVE
    while lo < hi:
        mid = (lo + hi) // 2
        if a[mid] > a[hi]:
            lo = mid + 1
        else:
            hi = mid
    return a[lo]

0: Comparing a[mid] with the last element shows which side holds the rotation: if a[mid] > a[hi] the minimum is right of mid, else it is at mid or left. The loop converges on index 4, value 0. Note the invariant: hi is inclusive and hi = mid is safe because mid itself can be the answer. This version assumes distinct values and non-empty input; duplicates need additional handling.

What Microsoft says its technical interviews assess

Microsoft’s current technical-interview guidance covers problem solving, design, coding and testing. It asks candidates to clarify ambiguity and make a plan before implementation. It describes a short, 45-minute interview round in that technical guidance; the general hiring page describes most interviews as two to four conversations lasting up to an hour. These are differently scoped descriptions, not a promise that every engineering loop has the same count or duration.

For coding, Microsoft says to use the language you are strong in, not pseudocode. Its current technical page describes a third-party tool in which you can run and compile code. It does not name a universal vendor. Confirm the exact tool, language options and round format with your recruiter.

The technical topic list is representative and depends on the role. It includes algorithms, data structures, distributed systems, AI/ML, data science and networking. Do not treat all of these as mandatory subjects for every software engineering vacancy.

An original Microsoft preparation drill

The three questions below practice arrays, state and boundary testing. They are small reasoning exercises for preparation, not copies of Microsoft questions and not a timed Microsoft assessment. The optional timer is a practice aid only. Read the explanation after each choice, then implement the idea in your own language.

Turn the drill into a complete spoken solution

Clarify: What counts as an empty input? Can values repeat? What should the function return when there is no valid result?
State a plan: Describe the simplest correct approach before discussing a faster one.
Code: Use a real language and name variables so another person can follow the state.
Test: Cover the ordinary path, an empty input and a case that changes your current best answer.
Explain cost: State what n means, the amount of work as n grows, and any extra storage.

For a single pass through n values while keeping only a counter and current maximum, the work is O(n) and extra storage is O(1). That statement depends on the algorithm you actually implement. Do not claim constant space if you also build a new array of n values.

A Microsoft boundary-test worksheet

Before saying the function is done, choose a test that would disprove your current implementation. Ordinary examples show that one path works; a boundary example can expose the wrong assumption.

Input boundaries

Empty input; one item; values all equal; negative values when allowed; a result at the last position.

State boundaries

First update; reset after a break; a tie; a best value found early versus late; a no-result path.

Contract boundaries

Return a count, an index or a value? Must the original input stay unchanged? Is an error preferable to an empty result?

Not every boundary is relevant to every task. Tie each test to the specified contract rather than reciting a checklist that ignores the problem.

Use real experience for Microsoft technical discussions

Choose one project you can explain from requirement to testing. Make a compact map: the original need, the part you owned, two approaches you considered, why you chose one, how you tested it, and what remained uncertain. If the project was a prototype, call it a prototype. A successful local test is not evidence of production reliability.

Try a follow-up challenge

Suppose your first version scanned a list once for every query. What changes if there are many queries but the list rarely changes? Explain a possible preprocessing trade-off, including build cost, lookup cost, memory and how updates would work. This is an original reasoning extension, not a reported Microsoft interview prompt.

Use Microsoft’s broader hiring preparation route to build a truthful STAR(R) example with reflection.

A focused Microsoft software engineer practice plan

For your next session, solve one small problem aloud, code it without outside answer generation, and test three meaningful cases. Review one mistake before moving to another topic. In a second session, explain one real project and ask a partner to change an assumption. For senior or specialist roles, select design or domain work from the job description instead of treating generic arrays as complete preparation.

Microsoft’s Candidate Code of Conduct separates preparation from live assessment: outside assistance during interviews or assessments is not allowed unless explicitly permitted. Use practice tools now; follow the actual interview rules later.

Microsoft software engineer questions

Should I use pseudocode in the interview?

The current technical guidance says not to use pseudocode and to code in the language you are strong in. Confirm the available language and tooling before the round.

Must I study every topic on the technical page?

No. Microsoft labels its technical-excellence list representative and role-dependent. Start from your vacancy and recruiter guidance, then fill gaps.

Do these drills predict a Microsoft interview result?

No. They give feedback on these original exercises only. They are not calibrated, validated or affiliated with Microsoft.

Choose your next practice step

Free reasoning practice Read an assessment invitation Prepare your online assessment setup

Microsoft hiring process and preparation

Compare other tech-employer preparation

Google hiring preparation Amazon hiring preparation Atlassian engineering interviews Meta hiring preparation NVIDIA hiring preparation

Sources and how to use this guide

Employer process statements come from the official sources above. The exercises, examples, practice plans and tools are original Prepti teaching, not employer questions or scoring criteria. Your vacancy, invitation and recruiter determine the actual interview requirements.

By Prepti Team. Last checked: November 9, 2026. Independent preparation; not affiliated with Microsoft.

Executable boundary challenge: rotated search with duplicates

Original follow-up: run the earlier rotated-minimum function on [1, 1, 1, 0, 1]. Its distinct-values assumption no longer holds. Trace the first comparison before reading the answer.

Reveal the failure and a bounded repair

The original code compares the middle 1 with the final 1 and keeps the left side, eventually returning 1. It misses the true minimum, 0. Equality does not identify which side contains the pivot when duplicates exist.

One duplicate-tolerant variant moves left when the middle value is smaller than the rightmost value, moves right when larger, and decrements the right bound on equality. That preserves a possible minimum but can reduce the interval by only one each step. Worst-case time becomes O(n), such as a long run of equal values. Do not keep promising O(log n) for every input.

Retest all-equal input, a single element, a non-rotated list and the counterexample. Keep the non-empty-input contract or define an explicit empty-input response. Explain how your cost claim changed with the contract, rather than treating the patch as free.

Compare evaluation and leakage reasoning

Original Prepti practice extension, added November 11, 2026. Fictional inputs, not employer questions or scoring rules.