登录
  • #面试经验
  • #数科面经
  • #分析|数据科学类
  • #instacart

Instacart Senior Data Scientist 電面x2+昂賽

Bakltube
4772
29
废话不多说,直接上时间线和面试题目。昂赛有签约,讲个大概。心得是,这家面试满硬的,比面 FB DS 感觉难多了。昂赛很累,还要现场花三个小时做 project + present。AB testing 也问得很细,没实际经验真的不太够应付,会问各种细节。希望这份面经可以帮到有需要的人们。

时间线:20191115 猎头 > 20191122 人资 > 20191210 Phone1: Product Sense > 20191217 Phone2: Stats + SQL > 20200110: Onsite

Phone1: Product Sense 45mins

各自简单自我介绍,马上进入问题。

1. instacart is aiming to deliver the order within 1 hour, some time slots are busier where demand is over supply. We launched an algorithm of busy pricing to charge more when the hour is busy. How do you evaluate the algorithm? (assume we didn't do AB testing before launch but we actually did), assume we don't incentivize shopper by shar

本帖隐藏的内容需要积分高于 188 才可浏览,点击前往一亩三分地论坛阅读

counter metrics of $ spent (-8 ~ +1) baseline ($50). How do you say about rollout?

4. Despite it's not statistically significant, what would you do to check the trend of declining $ spend?

SQL 满简单,查一下以前面经有类似题目,基本上看懂题目就会写

面过第二关,一寄出 SQL 答案隔五分钟就收到约 onsite,效率很高。

Onsite:

1. HR

2. Product sense: How to increase new users?

3. Project 1: Define customer value

4. Project 2: Find relevant factors

5. Project review: Present your work

6. Stats & experiment design: 以前出过的题目,给 shopper $100 来衡量 retention rate。要如何设计。这关超爆难,细节问很细。从 metrics design 到如何取样本到取样本会遇到的各种困难。

7. Product sense: How to improve Ads revenue? 这关遇到很好的面试官,算简单问题也不刁钻。

8. HR wrap up
29条回复
热度排序

发表回复