2020–2026 Microsoft CoreAI and AI/ML Intern OA / Interviews: 118 Experiences (Part 5 :)

This part preserves 13 Microsoft AI/ML internship reports from 2021–2022.

Coverage: 2021–2022 · candidate reports 050–062 · phone screens and final loops

Series note: Part 1 focuses on the CoreAI / AI-ML Software Engineering Intern loop. This part preserves other Microsoft AI/ML internship reports without treating them as CoreAI SWE evidence.

2022 reports (continued)

050: Data & Applied Science Intern — final interview loop

Report date: 2022-01-26 · Role: Data & Applied Science Intern · Stage: Final loop / onsite · Interview date: 2022-01-20

Microsoft Data & Applied Science internship Hiring Event report centered on delayed results. The author states that four rounds covered statistics, ML, behavioral questions, experiment design, and a final database/SQL exercise, but publishes none of the individual questions or schemas.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Unnamed Statistics, ML, Experiment-Design, and Database Questions.

2021 reports

051: Data Scientist Intern — phone or technical screen

Report date: 2021-12-15 · Role: Data Scientist Intern · Stage: Phone / technical screen · Interview date: 2021-12-06

Detailed Microsoft Data Scientist internship phone-screen pass report.

1. Refactor a Regex-Based Noun Pluralization Function. Explain an existing Python function that uses regular expressions to convert nouns to plural forms. Refactor or structurally improve that function. The report did not preserve a reconstructable task contract.

Other reported topics or limits: LightGBM Tuning, Early Stopping, and Explanation Questions.

052: Data & Applied Scientist Intern — phone or technical screen

Report date: 2021-11-30 · Role: Data & Applied Scientist Intern · Stage: Phone / technical screen · Interview date: 2021-11-30

Microsoft Data & Applied Scientist internship recruiter-screen report. The twenty-minute call contained only behavioral, project, motivation, team-preference, and ML/DL-interest questions, with no technical exercise or later-round update.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Leadership, Project, Motivation, and ML-Interest Screen.

053: Data Science / Applied Science Intern — phone or technical screen

Report date: 2021-11-29 · Role: Data Science / Applied Science Intern · Stage: Phone / technical screen · Interview date: 2021-11-29

Microsoft Data Science / Applied Science internship recruiter-screen report. The 30-minute first round asked five behavioral and research-communication questions; the author later received a virtual-onsite invitation but declined it after accepting another offer.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Behavioral and Research Communication Screen.

054: Microsoft Research AI Intern — final interview loop

Report date: 2021-11-23 · Role: Microsoft Research AI Intern · Stage: Final loop / onsite · Interview date: 2021-11-22

Microsoft Research AI summer-intern report. A manager project-direction chat led to two 30-minute interviews about the candidate's prior papers, detailed follow-ups, internship expectations, and preferred research topics.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Paper Follow-Ups and Research Preferences.

055: Data & Applied Science Intern — phone or technical screen

Report date: 2021-11-22 · Role: Data & Applied Science Intern · Stage: Phone / technical screen

Microsoft Data & Applied Science internship recruiter-screen report. The shortened holiday-season call covered pending offers, motivation for technology, a challenging data-science project, dataset size, leadership versus team-member role, a prior internship, motivation for Microsoft, preferred Microsoft area and product, and familiarity with deep learning. The author later reported a final-round invitation but no final-round details.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Project, Dataset, Team-Role, and Motivation Screen.

056: Data & Applied Scientist Intern — final interview loop

Report date: 2021-11-18 · Role: Data & Applied Scientist Intern · Stage: Final loop / onsite · Interview date: 2021-10-28

Detailed Microsoft Data & Applied Scientist internship pass report.

1. Implement a Precision-Recall Curve. The inputs are a prediction array and a label array. Compute the precision and recall arrays that form the precision-recall curve.
2. Longest Increasing Subsequence. The interviewer used LeetCode 300 exactly. The candidate wrote a dynamic-programming solution; the interviewer did not require the O(n log n) follow-up.
3. Enumerate Unique Unordered Dice Outcomes. A die has k faces and is rolled n times. Output every possible result without duplicates. Outcomes such as (1,2) and (2,1) are considered the same. DFS met the initial expectation; larger n and k caused stack overflow and prompted a dynamic-programming follow-up.

Other reported topics or limits: Predict Windows Upgrade Duration; Satellite Cloud Detection and ML Fundamentals.

057: Data & Applied Scientist PhD Intern — phone or technical screen

Report date: 2021-11-11 · Role: Data & Applied Scientist PhD Intern · Stage: Phone / technical screen · Interview date: 2021-09-15

Detailed Microsoft Data & Applied Scientist PhD internship offer report.

1. Percentage of Observed Users Among Installed Users by App. Table 1 has columns app and installed_users. Table 2 has columns app, userId, and date. For each app, compute the percentage of users represented in Table 2 among all installed users for that app. The candidate wrote SQL in a shared Word window during the Teams call.

Other reported topics or limits: High-Dimensional ML, Model Comparison, and Product Experimentation Questions; Unnamed Onsite ML Case Studies and Coding Questions.

058: Data & Applied Science / Data Science Intern — final interview loop

Report date: 2021-11-09 · Role: Data & Applied Science / Data Science Intern · Stage: Final loop / onsite · Interview date: 2021-10-28

Microsoft Data & Applied Science internship offer timeline. The Azure virtual onsite had two coding-plus-behavioral rounds, a statistics and machine-learning fundamentals round, and a final behavioral-plus-simple-coding round. The author characterizes coding as ordinary algorithms/data structures around medium difficulty but discloses no individual task identity.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Unnamed Azure Algorithms, Data Structures, Statistics, and ML Questions.

059: Data Science PhD Intern — final interview loop

Report date: 2021-11-07 · Role: Data Science PhD Intern · Stage: Final loop / onsite · Interview date: 2021-11-02

Microsoft Data Science PhD internship final-round report.

1. Design Related Query Suggestions for a Search Engine. A search engine should return relevant search suggestions for a user's query. Propose an algorithm and explain its details.
2. Design Reinforcement Learning for Topic Extraction. Explain how reinforcement learning could be applied to topic extraction.

Other reported topics or limits: Four Unnamed Easy-to-Medium Onsite Coding Tasks.

060: Data Scientist Intern — final interview loop

Report date: 2021-11-04 · Role: Data Scientist Intern · Stage: Final loop / onsite · Interview date: 2021-11-01

Detailed Microsoft Data Scientist internship onsite report.

1. Predict a Car's Price from Similar Cars. Write code to predict car prices. The candidate used nearest neighbors and averaged their prices.
2. List All Files in a Folder and Find Duplicate Files. Print every file in a folder. Follow-up: find files that are the same. Follow-up: optimize the duplicate-file search.
3. Two-Player Gambler's Ruin Probability and Simulation. Two people gamble against a machine. A win changes wealth by plus one and a loss by minus one. Both people start with one unit of wealth. Compute the probability that both eventually lose all wealth. After deriving the probability, write and run a simulation.

Other reported topics or limits: Predict Unspecified User Demand; Unnamed General DS Knowledge and Table-Merge Questions.

061: Data & Applied Scientist PhD Intern — final interview loop

Report date: 2021-11-04 · Role: Data & Applied Scientist PhD Intern · Stage: Final loop / onsite · Interview date: 2021-10-28

Microsoft Data & Applied Scientist PhD internship offer report. This atypical four-round onsite consisted almost entirely of behavioral conversation and deep discussion of one candidate-specific project, including packages, storage, platform, and design rationale. The author confirms no coding round and no reusable standalone question.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Candidate-Specific End-to-End Project Deep Dive.

062: Data & Applied Science Intern / Data Scientist Intern — final interview loop

Report date: 2021-11-03 · Role: Data & Applied Science Intern / Data Scientist Intern · Stage: Final loop / onsite · Interview date: 2021-09-30

Microsoft Data & Applied Science internship pass report that transferred to Data Scientist Intern. It preserves the complete recruiting timeline and states that the first round covered resume projects, basic ML knowledge, and behavioral questions, but intentionally withholds specific technical prompts.

1. No standalone technical prompt was preserved. This entry is retained for its process, timing, role, or interview-format evidence.

Other reported topics or limits: Withheld ML, Resume-Project, and Behavioral Interview Content.

Part 6 continues with the next candidate reports.