Собеседование аналитика в Яндекс Такси
Карьерник — квиз-тренажёр в Telegram с 1500+ вопросами для собесов аналитика. SQL, Python, A/B, метрики. Бесплатно.
О компании
Яндекс Такси (Yandex Go) — крупнейший ride-hailing в России и СНГ. Two-sided: users + drivers. Complex: supply / demand dynamics, pricing, logistics.
Analytics команды — разные: pricing, driver, product, growth, marketing.
Процесс
Типичный Yandex-style:
- HR screening
- Technical (SQL + stats)
- Product / case
- Behavioral
- Final с team lead
2-4 недели process.
Стек
- SQL: ClickHouse heavy, YT, PostgreSQL
- Python: pandas, sklearn
- BI: DataLens, internal tools
- Big data: YT (Yandex's Hadoop-equivalent)
SQL раунд
Yandex — сильные SQL. Ожидайте:
- Advanced window functions
- Optimization tasks
- Complex JOINs
- Array / JSON в ClickHouse
Example
«Для каждого водителя — middle response time на запрос, отсортировать top-10 slowest».
SELECT
driver_id,
AVG(EXTRACT(EPOCH FROM (accepted_at - requested_at))) AS avg_response_sec
FROM ride_requests
WHERE accepted_at IS NOT NULL
AND requested_at >= NOW() - INTERVAL 30 DAY
GROUP BY driver_id
ORDER BY avg_response_sec DESC
LIMIT 10;Product / case
Two-sided problems
«Surge pricing — ok или нет?»
Trade-offs:
- Higher supply attraction (drivers)
- Lower demand (users may cancel)
- Revenue optimization
- User satisfaction
Discuss framework.
Typical cases
- «Cancellations grew 20%. Why?»
- «Launched в new city — metric success?»
- «Price change impact — measure?»
- «Driver churn causes»
Marketplace thinking
- Liquidity: matches happen fast
- Supply/demand: balance
- Network effects: more drivers → less wait → more users → more drivers
Знайте dynamics.
Metrics
Specific к ride-hailing:
- Matches per request
- ETA (estimated time to arrival)
- Cancellation rate (user, driver)
- Driver utilization (% time busy)
- Revenue per ride
- Commission take
- Driver acquisition cost
- User CR (open app → ride)
Geo aspects
Geographic analysis essential:
- Heat maps demand
- Supply gaps
- Route efficiency
- City-level comparisons
Pricing
- Dynamic surge
- Base fare
- Per-minute / per-km
- Peak pricing
Model / simulate price changes.
Driver перspective
- Earnings stability
- Acceptance rates
- Cancellation policies
- Promotions effectiveness
На собесе
Behavioral (Yandex style)
STAR истории:
- Complex problem solved
- Cross-functional leadership
- Failed project lesson
- Technical decision
Product case
Framework + quantified answer.
Math / stats
- Poisson processes (arrivals)
- A/B design challenges (network effects)
- Time series (seasonality)
Зарплата
Yandex top-market:
- Junior: 130-180k ₽
- Middle: 200-300k ₽
- Senior: 300-450k ₽
- Lead: 450-700k ₽
Plus bonus, RSU.
Как подготовиться
SQL
Strong. Window functions, optimization.
Python
pandas, scipy, sklearn. Pandas performance.
A/B
Network effects, cluster-randomized.
Product
Marketplaces dynamics. Two-sided metrics.
Russian mobility
Read blogs: Yandex Tech, transportation research, ride-hailing news.
Related Yandex companies
- Yandex Go (Такси / Лавка)
- Yandex.Eда (food delivery)
- Yandex.Доставка (logistics)
- Яндекс Драйв (carsharing)
Похожие скилы, разные vertical.
На собесе распространённый question
«A/B test в ride-hailing tricky. Почему?»
Answer: network effects. Treatment driver's improved matching affects nearby control drivers (less supply for them).
Solutions: geo-cluster randomization, switchback.
Private trends
Growth areas:
- Grocery delivery integration
- Subscriptions (Plus)
- Business accounts
- International expansion (ex-CIS)
Связанные темы
FAQ
Russian only?
Mostly Russia. Some Kazakhstan, Uzbekistan.
Remote?
Hybrid часто.
Опыт transportation нужен?
No. Analytics mindset важен. Domain — learnable.
Готовьтесь — откройте тренажёр с 1500+ вопросами для собесов.