takentui

Research · Python backend

Interviews.
Let’s look
at the data.

See what interviewers ask and what to practise. Explore topics from real interviews, linked to our competency matrix.

Explore topics
Interview outcomesSnapshot / 24 Sept 2026
335interviews
in the archive
51 Passed23 Did not pass3 Process stopped258 Outcome unknown

Each square is one interview. Outcomes come from the Telegram archive index, recording captions and feedback. “Passed” includes invitations to the next stage and job offers.

For 258 records the outcome is unknown. In 3 cases, the selection was stopped: the vacancy was filled or the salary could not be agreed upon.

4760technical answers assessed
14.2assessed technical questions per interview on average
2515answers with mistakes · 52.8%
Where the numbers come from
and what they mean

[ 01 / Topic map ]

Where to start preparing?

Choose a topic. Behind each bar is the number of interviews in which it was tested.

Share of 335 interviews · one topic may appear together with others

0 → 100%

Showing 12 of 160 topics

[ 02 / From observation to skill ]

Let the data
guide your practice.

01

Find a topic

Compare topic frequency with the requirements of the role you want. Use the job description to guide your preparation.

02

Explain the solution

Answer the question without notes. Give an example and state the limits of applicability.

03

Check with code

Jump to the material or activity. Support your explanation with a working result.

[ 03 / Methodology ]

How we
counted.

What do the percentages next to the topics mean?

How often has this topic been asked in our archive? 30% - approximately 30 job interviews out of 100. If at one job interview they asked five questions about indexes, we count it once.

During the interview, several topics are usually checked, so the sum of the percentages will be more than 100%.

Where does the data come from?

Collected 335 recordings of public and private interviews, mainly on Python and backend. One candidate could undergo several interviews.

Dates are known for 271 records, including 79 from 2026. The last known interview date is 23 Apr 2026. Counts updated 24 Sept 2026.

How did you find the questions?

AI reviewed all 335 transcripts in full. A standalone technical question and its follow-ups counted as one episode under one primary topic. Discussions of experience without a knowledge check were skipped.

Topic frequency counts interviews: several questions on the same topic in one interview count once. This gives 2843 interview–topic pairs and 4760 assessed answers. The difference comes from asking multiple questions on a topic.

How did we count mistakes?

We found 5464 technical episodes: 4760 answers were assessed and 704 were excluded when speech, code, or the question itself was missing.

A miss is an incorrect, incomplete, or absent answer, even if the interviewer says “okay” and moves on. A self-corrected slip or a pause does not count.

For example, 2 misses among 4 assessed answers gives 50%. Both counts are shown beside the percentage.

AI assessed the transcript answers. A person has not independently checked the ratings yet. Each rating concerns one question.

Is this representative of the entire market?

No. But these are observations from actual recordings of interviews—public and private. The archive shows trends: what is asked more often and what topics are paid attention to.

We wrote the questions on these cards for self-assessment. They do not quote candidates.

The topics build on a backend developer competency matrix reflecting typical market requirements, with additional topics found in the interviews.

Misses and topic frequency were calculated from the same 335 recordings.