Microsoft runs a fairly uniform loop across its engineering orgs: a recruiter screen, a timed online assessment (Codility or HackerRank is what candidates name most often), then a same-day virtual block of three to six 45-60 minute interviews. The technical rounds lean hard on data structures and algorithms with an explicit push from brute force to an optimized solution, then split into a low-level/object design round and a high-level system design round for anything above entry level, closing with a hiring-manager conversation and an 'AA' (As Appropriate) round run by a senior person outside the team who effectively holds the hire/no-hire decision. Non-engineering tracks reshape the middle: consulting and support candidates get CS-fundamentals breadth plus Azure and customer-scenario questions instead of LeetCode grinding, data-science candidates get open-ended modeling case studies, and product managers get product-sense and program-management interviews alongside an engineer-run technical round.
These 23 writeups cover software engineering, product management, data & science and other roles. 16 of them were posted under a pseudonym (Reddit, GeeksforGeeks or forum handles), so we could not verify the authors’ identities.
Typical rounds
41-6 range
Outcomes shared
13/5offer / not
Most common round
Technical
Sources span
2012 - 2026
Most frequently reported · Data structures & algorithms(22), Behavioral(14), Past projects(13), Arrays & strings(11), OOP(11)
Microsoft’s official process
Microsoft's own hiring-tips pages say the interview steps vary by role and can include meetings with prospective teammates and cross-functional colleagues, coding exercises, portfolio reviews or work-sample presentations, and that the process is currently conducted virtually. The company tells candidates to structure answers with its STAR(R) model (Situation, Task, Action, Result, Reflection), to ask clarifying questions and narrate their reasoning rather than just produce answers, and to show 'growth mindset' and resourcefulness when they hit something they do not know instead of bluffing. A separate student-interviewing page describes a shorter two-stage flow (an initial interview, then a final interview with a few interviewers of up to an hour each), and the hiring FAQ states applications are reviewed by recruiting before being surfaced to the hiring manager, screeners and interviewers.
A short campus coding test with two questions that the author found so easy the whole cohort finished quickly, which meant the initial shortlist cut deeply and they only made a later, extended list.
Covered · Data structures & algorithms
2
Technical interview 1Technical
Held over Teams: an introduction, then a single implementation-heavy problem converting a number into its English words form. The author underestimated the edge cases, overran the time limit, and the interviewer ended up helping debug a missing entry in a lookup table.
Covered · Arrays & strings, Debugging
3
Technical interview 2Technical
A matrix in-place modification problem taken from naive to constant-extra-space with hints, then a verbal-only design of an LRU cache where the interviewer pushed back on a fixed-size array implementation and steered toward a linked-list based one.
Run by a team lead from the Edge PDF viewer team. Started with hobbies, coursework and a deep pass over the author's open-source and independent projects, then a time-boxed scheduling problem (finding a common meeting slot across N calendars) that the author only partly nailed.
Covered · Past projects, Data structures & algorithms
A LinkedIn-sourced recruiter call covering career trajectory, notice period and an explanation of the SWE loop.
Covered · Resume deep-dive
2
Online assessmentOnline assessment
Two easy-to-medium problems, one string manipulation and one framed as a game but reducible to backtracking over permutations. Only sample cases are visible, so edge cases have to be reasoned about rather than tested.
Covered · Arrays & strings, Data structures & algorithms
3
Technical interview 1Technical
Opened with work-experience discussion, then OS and database-normalization theory, then two coding problems (a matrix inversion-style problem and generating balanced parentheses), each taken from brute force to a space-optimized version.
Covered · Operating systems, DBMS, Arrays & strings, Past projects
4
Technical interview 2Low-level design
Cloud services and Azure/AWS integration discussion, design-pattern questions (singleton, builder), then a scenario-based approvals system where the interviewer kept layering on constraints and expected SOLID-conformant object-oriented code.
Covered · OOP, Low-level design, Domain knowledge
5
Technical interview 3Technical
More core-subject questions plus a classic counting-out puzzle that the interviewer wanted solved with a closed-form constant-time answer rather than the circular-list simulation the author started with.
Covered · DBMS, Networking, Data structures & algorithms
6
Techno-managerial roundHiring manager
With the hiring manager: a shortest-transformation-sequence graph problem where the point was justifying BFS, followed by a run of managerial questions.
A 90-minute Codility test with three problems (a single-pass parentheses balancing question, a substring question, and a harder matrix DP), with roughly half the total needed to advance.
Covered · Arrays & strings, Dynamic programming
2
Technical interview 1Technical
One problem on a binary search tree with two swapped nodes; the interviewer explicitly wanted the brute force stated first and then an optimized traversal-based solution.
Covered · Trees, Data structures & algorithms
3
Technical interview 2Technical
Two harder problems on a shared Codility link: enumerating valid IP-address splits of a string, and a min-cost partitioning problem solvable by sorting on a cost differential with a custom comparator.
Covered · Arrays & strings, Data structures & algorithms
4
Project manager roundSystem design
A URL-shortener design scaled to the candidate's experience level, with the interviewer helping with requirements gathering and throughput estimation.
Covered · System design, Scalability, API design
5
Engineering manager roundTechnical
A single recursion/combinatorics problem: enumerate all length-N tuples summing to S, with repetition allowed.
Covered · Data structures & algorithms
6
AA roundFinal
The As-Appropriate closer: an in-place array rearrangement (parity to matching index parity) with no extra space, then the interviewer hammering the solution against corner cases.
Two easy Codility questions, one a debug-the-given-code task and one write-from-scratch.
Covered · Debugging, Data structures & algorithms
2
Technical interview 1Technical
Two problems on the same day block: designing a calendar slot-booking API where overlapping intervals had to be merged and a suitable time type defined, then a binary-tree question about nodes at a fixed distance from leaves.
Covered · API design, Arrays & strings, Trees
3
Technical interview 2Low-level design
A short project discussion followed by designing a text-editor component with insert, delete, copy, paste and bounded undo/redo, expressed as a class diagram with composition rather than as an algorithm.
Covered · Low-level design, OOP, Past projects
4
Technical interview 3Technical
Technical discussion of the candidate's project stack, then coding level-order and skewed traversals of a binary tree, then being pushed to reduce the auxiliary space.
Covered · Trees, Data structures & algorithms
5
Managerial roundHiring manager
Mostly HR-style questions with two small coding problems mixed in (maximum contiguous subarray sum, and printing duplicate elements in reverse order of appearance).
On Teams with a Codility screen share: build a binary tree from scratch (node class included) and solve a longest-consecutive-path question, then handle the corner cases the interviewer added afterwards.
Covered · Trees, Data structures & algorithms
2
DSA roundTechnical
CV questions, then an interval/greedy problem (minimum simultaneous resources needed given arrival and departure times). The candidate's solution was time-bounded rather than optimal; the interviewer nudged toward the greedy answer and then accepted the coded suboptimal version.
Covered · Arrays & strings, Resume deep-dive
3
Design roundLow-level design
With a director-level interviewer who was hostile about the candidate using Teams in a browser and burned 15 minutes on it. The actual question was an object design for a snake game, moving on to which resources threads would share.
Two problems: a non-adjacent maximum subsequence sum (a classic DP) and implementing a stack over a plain array with push, pop and peek.
Covered · Dynamic programming, Data structures & algorithms
2
LLD roundLow-level design
Design a parking-lot system with three vehicle/bay sizes, limited inventory per size, upgrade-to-next-size allocation on entry, and per-size hourly billing on exit. Working code was expected, not just a class diagram.
Covered · Low-level design, OOP
3
Hiring manager roundHiring manager
Resume-driven discussion of past projects plus a substantial behavioral block.
Covered · Resume deep-dive, Past projects, Behavioral
Design an iterator that merges n sorted lists behind hasNext/next, then a binary-tree spreading/propagation timing problem.
Covered · Data structures & algorithms, Trees, API design
2
DesignLow-level design
Design a logging system able to record several event categories.
Covered · Low-level design, OOP
3
DesignSystem design
Design an anagram-lookup service against a dictionary too large to hold in memory, then a data-flow design for incremental spell-checking over a paginated grid that streams from a server.
Covered · System design, Scalability, API design
4
DesignSystem design
A OneDrive-flavoured problem: list a user's own and shared files with a recently-accessed block on top and the remainder alphabetically.
Two problems needing better than 50% to pass: a variant of the multi-transaction stock-profit DP, and an arithmetic question on sequences of bits in a negative base.
Covered · Dynamic programming, Arrays & strings
2
DSA roundTechnical
Introductions and current-stack discussion, then repairing a broken pointer in a circular doubly linked list, then building a binary tree from a stream and printing it level by level.
Covered · Data structures & algorithms, Trees
3
Design roundSystem design
Ran roughly 90 minutes against a 60-minute slot: design an alarm service with create/update/delete plus a scheduled delivery mechanism, giving both the high-level architecture and the CRUD API surface.
Covered · System design, API design, Scalability
4
Hiring manager roundHiring manager
About thirty minutes of database questions (SQL versus NoSQL trade-offs, picking a store for a payments use case, schema evolution on a large table, a ranked-employee query) and then thirty minutes of behavioral and scenario questions.
70 minutes, two tasks: a basic array problem and a debugging task where at most two lines could be added to fix a broken breadth-first search. The recruiter set the bar at roughly 80% of hidden test cases across both.
Covered · Arrays & strings, Debugging
2
DSA round 1Technical
With a Redmond-based engineer: background and current project, then two problems (a stack-simulation collision problem and a linked-list splice) both requiring the candidate to write test cases and actually run the code.
Covered · Data structures & algorithms, Arrays & strings
3
DSA round 2Technical
Resume questions, then a weighted shortest-path-with-hop-limit variant, with the interviewer interleaving language-level follow-ups while the candidate coded.
Covered · Graphs, Data structures & algorithms
4
Low-level designLow-level design
Build a filtering layer over stored transactions by customer and time window, structured around interfaces and classes, then extended to a thread-safe implementation with locking and exercised with simulated concurrent calls.
Half the round was a challenge-driven interrogation of the candidate's hardest previous project and why each design decision was made, then designing a cloud logging service end to end on a shared digital whiteboard. Ran about 80 minutes.
Covered · System design, Scalability, Past projects
6
Hiring manager roundHiring manager
With a very senior manager who skipped coding entirely and instead ran an open estimation question about what determines the physical device count in a data centre, steering the candidate away from a traffic/load framing toward physical constraints.
90 minutes with a hard two-dimensional DP problem dressed up as a story and a medium string-backtracking problem.
Covered · Dynamic programming, Arrays & strings
2
Coding roundTechnical
Design and implement a simplified file system supporting create, read, list, rename, delete and search, then discuss how updating and persisting files would change in a distributed setting.
Covered · Low-level design, Data structures & algorithms, API design
3
Coding and design roundLow-level design
Implement a cache with LRU eviction, then refactor so the eviction policy is pluggable, then extend the discussion to a distributed deployment covering concurrent updates and consistency trade-offs.
Presented the architecture of a recent project (datastore choice, queueing, availability monitoring), then a parallelization exercise splitting a workload across cores, then a constrained large-array search problem.
Covered · System design, Concurrency, Past projects
5
AA roundFinal
A further project walkthrough judged mostly on communication, a stack problem with a harder follow-up taken from brute force to optimized with complexity analysis, and behavioral questions on teamwork and conflict.
Covered · Data structures & algorithms, Behavioral, Past projects
Two medium problems on HackerRank, one greedy/bit-manipulation flavoured and one depth-first search. Results took about two weeks, after which all remaining rounds were scheduled for a single virtual day.
Covered · Graphs, Data structures & algorithms
2
DSA roundTechnical
One interval-scheduling problem with a follow-up; the interviewer asked the candidate to code in the language shown on their resume rather than their preferred one.
Covered · Data structures & algorithms, Arrays & strings
3
Low-level designLow-level design
Object model for a music streaming app: classes, interfaces and enums, with pseudocode rather than compiling code expected.
Covered · Low-level design, OOP
4
High-level designSystem design
A URL shortener taken deep into multi-region deployment and unique ID generation, comparing snowflake-style generators, a dedicated ID service, and base-62 encoding of auto-increment IDs.
Covered · System design, Scalability, API design
5
Hiring manager roundHiring manager
Rescheduled at the last minute to a different interviewer. A domain question on making a checkout path idempotent; the candidate proposed a unique-index-backed idempotency key with a cache in front, and the interviewer rejected it aggressively and shifted the expectations mid-round. This was the round that sank the loop.
Covered · Domain knowledge, API design, Concurrency
A 45-minute virtual round weighted toward behavioral questions about taking initiative in a previous role, with a straightforward algorithm problem folded in.
Covered · Behavioral, Data structures & algorithms
2
Final-round interview 2System design
A 45-minute design conversation around an A/B testing component, judged on how methodically the candidate scoped the problem, generated alternatives and revised the design as new requirements appeared rather than on prior systems knowledge.
Software Development Engineer Intern (Windows & Windows Live)
Internship
Received an offernot disclosed; author describes it as approaching entry-level full-time territory with housing subsidised5 rounds
1
Campus interviewTechnical
An on-campus first round after a career-fair application, centred on whiteboard coding and being able to talk with genuine enthusiasm about a personal project. The author notes campus recruiters do their own resume screen and effectively make recommendations, so speaking to them matters.
Covered · Data structures & algorithms, Past projects
2
Redmond onsite interview 1Technical
First of four back-to-back onsite interviews at Redmond, this one with a Visual Studio engineer. The author missed the intended solution badly, and later learned the same question tripped up most candidates.
Covered · Data structures & algorithms
3
Redmond onsite interview 2Technical
With an Internet Explorer engineer, on a topic outside the candidate's area. The implementation broke on one of three required cases because it violated an assumption baked in at the start.
Covered · Data structures & algorithms, Debugging
4
Redmond onsite interview 3Technical
With a Windows kernel/filesystem engineer whose style was to pick terms off the resume and ask the candidate to explain each one; Linux-based answers were accepted without friction.
Covered · Operating systems, Resume deep-dive
5
Redmond onsite interview 4Technical
With the prospective manager from Sustained Engineering. The candidate found the intended solution but introduced a bug by using a clever branchless bit trick, which they reflect was exactly the wrong instinct in front of a future maintainer.
A scheduled call with a university recruiter, opening with standard behavioral questions and then a weighing-problem brain teaser. The candidate recognised the puzzle from a well-known prep book, said so rather than faking a live solve, and was given a substitute puzzle they could not crack under time pressure. A rejection email arrived about a month later.
Covered · Behavioral, Puzzles, Data structures & algorithms
Four separate interviewers, each posing one on-the-spot programming question, deliberately ramping in difficulty. The first was a buffer-to-buffer rectangle copy in C given source pitch, a source rectangle and a destination origin.
Covered · Arrays & strings, Data structures & algorithms
2
Interviewer 2Technical
A null-terminated string copy, followed by modifications the author, with hindsight, questions the soundness of.
Covered · Arrays & strings, Debugging
3
Interviewer 3Technical
Drawn from a real product's flood-fill implementation: write the colour-match test for a packed-pixel format where four pixels share a byte, so a plain equality comparison does not work. The author could not solve it in the room.
Covered · Data structures & algorithms, Debugging
4
Interviewer 4Technical
The hardest of the four, continuing the escalating-difficulty pattern; two of the four questions were explicitly about performance.
An initial screening call with a recruiter before any onsite scheduling.
Covered · Resume deep-dive, Behavioral
2
Onsite loop (Redmond)Panel
A six-interview onsite day in Redmond. Candidates are told only their first three interviewers up front; reaching the remainder depends on how the earlier ones go, which makes the loop a live elimination ladder.
Covered · Behavioral, Past projects, Data structures & algorithms
3
PM and engineer interviewsTechnical
The loop mixes software engineers, who run the technical assessment, with product managers, who probe both product sense and program-management ability.
Covered · System design, Past projects, Behavioral
4
AA roundFinal
The As Appropriate interviewer sits above the hiring manager and is the actual decision maker on the hire.
45 minutes, three open-ended modeling cases: turning a binary classifier's probability output into an ordinal prediction, deciding features and a suitable model architecture for paraphrase detection, and making a colour-query image retrieval system scale.
Covered · Case study, Domain knowledge, Scalability
2
Research and projects roundTechnical
40 minutes centred on the candidate's thesis (storage, scalability, how it could be improved, the indexing technique behind it) and a recent NLP project, ending on general questions about which data structures they were confident in. No further rounds were scheduled.
Covered · Past projects, DBMS, Dynamic programming, Data structures & algorithms
A multiple-choice test on an external campus platform covering basic machine learning and neural networks, plus aptitude and logical reasoning sections. Roughly 14 of about 100 advanced.
Covered · Domain knowledge, Aptitude
2
Group FlyTake-home
Microsoft's group written round: two coding questions solved on paper with working code plus unit tests, one on parsing nested function calls and one on finding the largest empty rectangle among points in a box.
Covered · Data structures & algorithms, Testing
3
Project interviewTechnical
With a member of the Bing team, a full walkthrough of the candidate's project including the underlying mathematics.
Covered · Past projects, Domain knowledge
4
Coding interviewTechnical
Two problems solved on paper where the approach mattered more than the code, plus a conceptual discussion of why neural networks work.
Covered · Data structures & algorithms, Domain knowledge
5
Mixed HR and technicalHiring manager
Motivation questions about the role and the machine-learning field, plus a simple interval-scheduling coding question written on paper.
Covered · Behavioral, Data structures & algorithms
A long technical round on .NET runtime internals (intermediate language, managed versus unmanaged code, garbage-collection generations, assemblies), web-server hosting concepts (application pools, ports, request lifecycle), TLS termination and connection pooling, ending on troubleshooting walkthroughs for a suddenly-failing app and session-expiry bugs.
More runtime internals plus a substantial Azure block: app services, custom domains and TLS, app service plans, resource groups, scaling and networking features, the service-model distinctions, DNS resolution and the TCP handshake, and a diagnose-the-hanging-app scenario.
Current role, understanding of what the support role at Microsoft actually is, automation the candidate has built and its impact, career plans and self-improvement, and how they behave when blocked on unfamiliar technology.
Covered · Behavioral, Resume deep-dive
4
Communication roundPanel
A customer-facing assessment: which regions' clients the candidate has worked with, how they would handle an irritated customer, strengths and weaknesses, and a live mock support call for a user who cannot send mail from an application.
No aptitude test that year; shortlisting was purely on resume and academic thresholds, cutting 43 eligible applicants to 10 at the author's college.
Covered · Resume deep-dive
2
Technical interviewTechnical
About 55 minutes of breadth rather than depth: databases, algorithms, big data, AI, cloud, web development, DevOps, software lifecycle, testing and every OOP concept, with basic coding where explaining and extending the logic mattered more than the code, plus project questions and occasional puzzles. Azure familiarity was a bonus.
Covered · DBMS, OOP, Data structures & algorithms, Past projects, Domain knowledge
3
Hiring manager roundHiring manager
An hour with a manager mixing technical questions (AI, cloud, databases), detailed project and internship interrogation including teammates' contributions, situational behavioral questions answered in STAR form, guesstimates and puzzles.
Covered · STAR stories, Case study, Puzzles, Past projects, Domain knowledge
4
AA roundFinal
A shorter 30-minute closer with a very senior interviewer, mostly about values and approach, including which Microsoft products the candidate uses and what they find distinctive about them. Two of four candidates were selected.
Four sectioned and separately-cutoff sections: aptitude, CS fundamentals as multiple choice, two easy-to-medium coding questions, and an essay. 22 of 90 eligible students advanced.
An hour or more of resume, internship and project discussion plus core-subject questions, easy coding in any language, software-lifecycle concepts and classic logic puzzles. Candidates were ranked against other colleges interviewing the same day.
Covered · Operating systems, DBMS, OOP, Puzzles, Past projects
3
Technical interview 2Technical
Highly variable: for most of the cohort a 15-20 minute situational/managerial round, but for this candidate a 50-minute technical one drilling into their deep-learning project and then into cloud service models and Azure, neither of which was on the resume.
Covered · Domain knowledge, Past projects, Behavioral
4
HR interviewBehavioral
A standard HR conversation, though this candidate also got cloud and OOP questions. Everyone who reached this round was hired.
Associate Consultant Intern (Microsoft India Global Delivery, on-campus)
Internship
Not selected3 rounds
1
Resume and CGPA shortlistTake-home
College placement office collected resumes; 25 CS/IT students were shortlisted. The nominal grade cutoff was lower than what shortlisted students actually had.
Covered · Resume deep-dive
2
Technical interviewTechnical
Coding done in a plain text editor over screen share. Pointer puzzles and output prediction, then the whole OOP surface (pillars with real-world examples, inheritance types, static versus dynamic binding, destructors) with small coding exercises, then merge sort pseudocode and linked-list implementations, then a long database block on ACID, sub-language categories, key types, delete versus truncate, joins, aggregation and a second-highest-salary query.
Covered · OOP, Data structures & algorithms, DBMS, SQL, Puzzles
3
Technical plus HRHiring manager
Resume projects with a JavaScript and database slant, then cloud service and deployment models because a cloud course was listed, then challenges faced while building projects and situational/behavioral questions. Eight students were selected; this candidate was not, and attributes it to the HR half rather than the technical half.
Covered · Past projects, DBMS, Domain knowledge, Behavioral
Microsoft runs a fairly uniform loop across its engineering orgs: a recruiter screen, a timed online assessment (Codility or HackerRank is what candidates name most often), then a same-day virtual block of three to six 45-60 minute interviews. The technical rounds lean hard on data structures and algorithms with an explicit push from brute force to an optimized solution, then split into a low-level/object design round and a high-level system design round for anything above entry level, closing with a hiring-manager conversation and an 'AA' (As Appropriate) round run by a senior person outside the team who effectively holds the hire/no-hire decision. Non-engineering tracks reshape the middle: consulting and support candidates get CS-fundamentals breadth plus Azure and customer-scenario questions instead of LeetCode grinding, data-science candidates get open-ended modeling case studies, and product managers get product-sense and program-management interviews alongside an engineer-run technical round.
What topics does Microsoft test in interviews?
Commonly reported topics include Data structures & algorithms, Trees, Graphs, Dynamic programming, Arrays & strings, Low-level design.
These guides summarize public, first-hand interview experiences shared by candidates. They describe the shape of each loop and the topics that came up, not a leaked question bank, and every experience links back to its original source. Processes change often; treat this as directional prep, not a script.