Lenovo Data Scientist interviews, as three candidates describe them, combine project deep dives, scenario-based machine learning questions, and hands-on Python, SQL or PyTorch work, though later rounds varied between manager discussions, a panel presentation and live coding.
$108K
Avg. Base Comp
$128K
Avg. Total Comp
4 rounds
Typical Rounds
3 months
Process Length
Across the three candidate reports, the most consistent theme is that Lenovo interviewers want you to explain your own work in detail. One candidate faced a resume deep dive on project theory and what they would redo. Another was pressed on which metrics improved and by how much. A third finished with a panel presentation of their best past project. Have a polished walkthrough covering your approach, obstacles and measurable impact.
The technical content differs from report to report. One candidate describes a pandas-style Python and SQL round (grouping, ranking, datetime handling) plus an open-ended propensity or stock-optimization question covering feature selection, model choice and validation. Another describes scenario-based ML discussions about choosing an algorithm for a business situation, with pressure on class imbalance and realistic predictions beyond accuracy. A third reports ML and computer vision fundamentals, a PyTorch round on custom modules, data loaders, training loops and dropout, and a live coding round on classic data structures and algorithms where time and space optimization mattered.
Outcomes differed: two candidates report offers and one reports no offer, with feedback pointing to deeper PyTorch detail and sharper DSA optimization. One candidate also describes an online multiple-choice AI/ML assessment and an HR phone screen before the technical rounds, and another had a technical interview conducted in Chinese, so it is worth asking your recruiter about format and language. One report describes a four-round process, and another says the whole process took about three months from application to offer.
Synthesized from 3 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Lenovo process.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Lenovo
Describing a data project and its challenges
| Question | |
|---|---|
| Uber Eats Customer Experience | |
| Client Solution Pushback | |
| Boosting Instagram Stories | |
| 2nd Highest Salary | |
| Upsell Transactions | |
| Prime to N | |
| Experiment Validity | |
| Find the Missing Number | |
| Random SQL Sample | |
| Paired Products | |
| Recurring Character | |
| Maximum Profit | |
| Retailer Data Warehouse | |
| Exam Scores | |
| Twenty Variants | |
| Equivalent Index | |
| Cumulative Sales Since Last Restocking | |
| Bucket Test Scores | |
| Completed Shipments | |
| Reducing Error Margin | |
| Detecting ECG Tachycardia Runs | |
| One Element Removed | |
| Cyclic Detection | |
| Size of Joins | |
| The Brackets Problem | |
| Valid Anagram | |
| Distribution of 2X - Y | |
| Google Maps Improvement | |
| Four Person Elevator |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report starting with an HR conversation. One describes a screener on work authorization and background. Another describes an online multiple-choice AI/ML assessment, then a phone call where HR briefed them on the role and process. Expect some early screening, though its exact form may vary.
Candidates report detailed walkthroughs of past projects. One was asked about timelines, responsibilities and which metrics improved and by how much. Another was asked about project theory, challenges in university work, how AI was used, and what they would change. Prepare specific, quantified stories.
Technical content differs by candidate. One reports scenario questions on algorithm choice, class imbalance and realistic predictions. Another reports bias-variance, loss functions, CNN convolution, evaluation metrics, and PyTorch tasks such as custom modules, training loops, autograd and dropout. A third got a propensity or stock-optimization question. Prepare to justify modeling choices.
Coding differs between reports. One candidate reports two SQL/Python questions leaning toward pandas grouping, ranking and datetime handling rather than algorithms. Another reports a live shared-editor session with classic DSA problems, including a kernel pattern matching question, where clarifying requirements and optimizing time and space mattered.
Later rounds reportedly shift toward fit. One candidate describes discussions with managers and leads about their projects. Another describes a final panel presentation of their best previous project, covering approach, obstacles and impact, and mentions having plenty of chances to ask questions back.
