T-Bank¶
English · Русский · AI Interview Atlas › Companies
Russian Big Tech · Russia · Loop reviewed 2026-09-26
Interviewers describe a screening, live coding and a team-fit interview. A 2025 senior candidate also had a management section and system design, and received a grade before meeting teams.
On this page: Interview loop · Questions (2)
Roles covered¶
- AI Engineering: Software engineer (general interview baseline), ML engineer / data scientist
- AI Leadership: Engineering manager
Interview loop¶
| Stage | What happens | Basis |
|---|---|---|
| HR introduction and screening | An introduction with HR, then a technical screening on theory. | ✅ confirmed by the company «Важно не только занять должность, но и работать на ней»: советы перед ИТ-собеседованиями, Т—Ж (издание Т-Банка), интервьюеры Т-Банка, published 2023-06-07, retrieved 2026-09-26 |
| Live coding | Programming in real time, about an hour and a half in one interviewer's description; the thinking process is assessed, not only the answer. | ✅ confirmed by the company «Важно не только занять должность, но и работать на ней»: советы перед ИТ-собеседованиями, Т—Ж (издание Т-Банка), интервьюеры Т-Банка, published 2023-06-07, retrieved 2026-09-26 |
| Team fit | A final interview to check fit with the team and its tasks. | ✅ confirmed by the company «Важно не только занять должность, но и работать на ней»: советы перед ИТ-собеседованиями, Т—Ж (издание Т-Банка), интервьюеры Т-Банка, published 2023-06-07, retrieved 2026-09-26 |
| Senior and team-lead track (2025) | A conversational management section, a two-hour language section and system design (a courier-tracking service), then fit interviews with several teams after a grade was assigned. | 🗣 candidate report Как я собеседовался в Ozon, Т-Банк, Mindbox и другие крупные компании, Хабр, пост кандидата, published 2025-07-09, retrieved 2026-09-26 |
| ML system design section ML engineer / data scientist |
Design and decompose a complex ML system: formalise the task and requirements, split it into subtasks, collect data, choose ML architectures, then deploy and test. Published in 2022 in a repository archived in 2023. | ✅ confirmed by the company Секция по дизайну ML систем, Т-Банк (Tinkoff), репозиторий career на GitHub, published 2022-08-01, retrieved 2026-09-29 |
| ML section ML engineer / data scientist |
Data analysis questions on theory and practical cases: framing the task, choosing and justifying quality metrics, collecting and validating data, and ML algorithms. From the same 2022 archive. | ✅ confirmed by the company Секция по ML, Т-Банк (Tinkoff), репозиторий career на GitHub, published 2022-08-01, retrieved 2026-09-29 |
Coding: required («Важно не только занять должность, но и работать на ней»: советы перед ИТ-собеседованиями, Т—Ж (издание Т-Банка), интервьюеры Т-Банка, published 2023-06-07, retrieved 2026-09-26)
Questions reported for T-Bank¶
Evaluation and observability¶
- Frame a business problem as a machine-learning task: choose and justify the quality metric, and explain how you would collect and validate the data.
- Knowledge · Asked at: T-Bank ✅
- Tests: Whether the metric and data follow from the business goal rather than habit.
- A strong answer covers:
- Translate the business decision into a prediction target, the action it triggers and the cost of each error.
- Pick the offline metric from those error costs, and name the online business metric it should move.
- Validate the data: label delay and leakage, time-based splits, and drift between training and serving.
- Read: Rules of Machine Learning (Google for Developers)
AI system design¶
- Design an ML system for a product feature end to end: frame the task and requirements, decompose it, plan data collection, choose model architectures, then deploy and test it.
- System design · Asked at: T-Bank ✅
- Tests: Whether you structure an ML system from problem framing to deployment instead of jumping to a model.
- A strong answer covers:
- Frame the product decision, offline and online metrics, and latency and cost limits before any model.
- Decompose into subtasks, plan where labels come from, and beat a heuristic baseline before adding models.
- Plan deployment through shadow mode or an A/B test, with drift monitoring and a rollback.
- Read: Rules of Machine Learning (Google for Developers)
← SberDevices (Sber) · Yandex →