Шпаргалка по собесу аналитика
Карьерник — квиз-тренажёр в Telegram с 1500+ вопросами для собесов аналитика. SQL, Python, A/B, метрики. Бесплатно.
Зачем это знать
One-stop reference для собеса. Открывайте перед каждым interview — quick refresh.
Стандартный process
Типичный
- HR screening (15-30 min)
- Technical (60-90 min): SQL, Python, stats
- Case / product (45-60 min)
- Behavioral (30-45 min)
- Final с team lead или head
Timeline
2-6 weeks total. Senior longer.
SQL часть
Must know
- SELECT, WHERE, GROUP BY, HAVING
- JOIN (INNER, LEFT, RIGHT, FULL)
- Subqueries, CTE
- Window functions (ROW_NUMBER, RANK, LAG, SUM OVER)
- Aggregations (COUNT, SUM, AVG, MIN, MAX)
- Date / String functions
- NULL handling
- EXISTS vs IN
- UNION / UNION ALL
- DISTINCT
Advanced
- Recursive CTE
- PIVOT (CASE WHEN)
- Median, percentile
- Performance (EXPLAIN, indexes)
- Complex JOINs
- Gap and islands patterns
Walk through process
- Clarify requirements
- Describe approach verbally
- Write SQL step-by-step
- Handle edge cases (NULL, dups)
- Test с sample data
- Consider optimization
Python / pandas
Must know
- pandas: read, filter, groupby, merge, pivot
- NumPy basics
- Dict, list, comprehensions
- Basic statistics (mean, median, etc.)
- Matplotlib / seaborn basic
Advanced
- Regex
- Feature engineering
- Memory optimization
- Basic ML (sklearn)
Statistics
Basics
- Distributions (normal, binomial, Poisson)
- Mean, median, mode, std
- Probability: conditional, Bayes
Testing
- Hypothesis testing (p-value, α, β, power)
- T-test, Chi-square, ANOVA
- Correlation (Pearson, Spearman)
A/B
- Sample size calculation
- MDE, effect size
- Multiple testing, peeking
- CUPED, variance reduction
ЦПТ, ЗБЧ
Central Limit Theorem, Law of Large Numbers — explain simply.
Product case
Framework
- Clarify question
- Structure approach
- Decompose problem
- Metrics to consider
- Hypotheses generate
- Prioritize actions
Common cases
- Метрика упала: RCA framework
- New feature launch: measurement plan
- Retention improvement: segmentation, drivers
- Pricing decision: elasticity, competitive
- Market sizing: Fermi estimation
Vocabulary
DAU, MAU, retention, churn, LTV, CAC, AOV, GMV, NPS, funnel, cohort, activation.
Behavioral
STAR method
- Situation
- Task
- Action
- Result
Top stories ready
- Most impactful project
- Failure / lesson
- Conflict resolution
- Ambiguous task
- Tech challenge
- Leadership moment
- Learning new skill
Common Qs
- Tell about yourself
- Why this company
- Strengths / weaknesses
- Disagreement with manager
- Why leaving current
- Where в 5 years
Preparation timeline
2 weeks before
- Review fundamentals
- Solve 5-10 SQL problems daily
- Read company (Glassdoor, blog, products)
- Prepare STAR stories
1 week before
- Mock interviews (peer / paid)
- Specific case practice
- Behavioral stories polish
Day before
- Review formulas, concepts
- Review resume (each line — discussable)
- Sleep well
Day of
- Light breakfast
- Arrive / login early
- Water beside
- Calm mindset
Во время interview
Opening
- Greet warmly
- Eye contact / smile (video — look at camera)
- Introduce briefly
During
- Listen fully
- Ask clarifying (key senior skill)
- Think out loud
- Manage time
Code
- Plan before writing
- Test mentally
- Handle edge cases
- Refactor после first pass
Case
- Structure first, execute second
- Quantify when possible
- Acknowledge assumptions
Behavioral
- STAR
- 2-3 min per story
- Not rambling
- Positive framing
Closing
- Ask 2-3 good questions
- Thank interviewers
- Ask about next steps
Red flags you
- No questions from you
- Rambling answers
- Defensive к feedback
- Негative про прошлое
- Wrong answers с confidence
- Не clarify
Green flags
- Ask questions
- Think before answer
- Acknowledge unknowns
- Structured reasoning
- Specific examples
- Engaged, curious
Online / Virtual
Setup
- Good internet
- Quiet room
- Lighting facing
- Camera above eye level
Technology
- Test Zoom / whatever
- Backup phone
- Know screen share
Etiquette
- Dress business-casual
- No eating / drinking heavily
- Close other tabs
- Pets contained
After
Thank-you
24h email к interviewers. Optional но nice.
Self-review
What went well? Improve?
Wait
Typical decision 1-2 weeks. Don't rush.
Rejected
Reflect, improve, next.
Offered
Negotiate, accept или counter.
Russian-specific
hh.ru
Main platform jobs.
Telegram chats
Analyst communities. Jobs often posted.
Habr Карьера
Tech jobs.
Зарплаты benchmarks
- Junior: 80-150k ₽
- Middle: 150-250k ₽
- Senior: 250-400k ₽
- Lead: 400-700k ₽
Big tech — upper bounds.
Company types
Big tech (Yandex, Тинькофф, Ozon)
- Competitive process
- Strong technical bar
- Good compensation
- Established culture
Mid-tier
- Moderate bar
- More learning
- Growing responsibilities
Startup
- Fast interview
- High responsibility
- Equity potential
- Risk / reward
Common mistakes
- Forget basics
- No STAR structure
- Too long / short answers
- No clarifying questions
- Fail listening
- Don't match interviewer style
Связанные темы
- Шпаргалка A/B
- Шпаргалка статистика
- Шпаргалка метрики продукта
- Шпаргалка ML
- 50 вопросов SQL
- Чеклист подготовки
FAQ
Сколько отказов normal?
Дo 10-20 applications → 5-10 interviews → 1-3 offers.
Best single resource?
Karьерник — quiz + guides в Russian.
Nervous?
Practice снижает. Nothing beats mock interviews.
Готовьтесь — откройте тренажёр с 1500+ вопросами для собесов.