Snowflake Company Profile
SnowflakeSnowflake, headquartered in California with 5,884 full-time employees, is a cloud computing startup. Snowflake provides cloud-based data storage and analytics services.Snowflake allows business users to store and analyze data using cloud-based hardware and software. Snowflake has been running on Amazon S3 since 2014 and on Microsoft Azure since 2018. Launched on Google Cloud Platform in 2019. Its Snowflake Data Exchange allows customers to discover, exchange and securely share data while maintaining a partnership with Snowflake. As such, Snowflake has an absolute say in technological innovation as a new-age cloud provider. So, even though AWS, Azure, Google Cloud, etc. are competitors with it, they are equally inseparable from its cooperation. In fact, Snowflake's biggest partner is Amazon AWS.
Snowflake Interview Timeline
April 10th: Posting at the school's Career Fair
April 21: OA received and completed
May 16: HR Phone
June 5: VO * 2
June 13: VO * 1
June 22: HM Telephone
June 26: Reference Check request
July 21: Offer Phone Call
Snowflake Interviews High Frequencyreal question
Snowflake Tech High Frequency Questions
- Snowflake vs. traditional databases?
why did you take the test?: The interviewer wants to know if you've figured out why organizations are migrating to Snowflake from older databases like Oracle and SQL Server.
high score answer::- Snowflake is a pure cloud-native, fully-managed platform that completely separates compute and storage for elastic scaling and cost control;
- Multi-cluster Architecture: Automatic horizontal scaling for high concurrency and guaranteed performance;
- Zero-copy cloning: Clone databases instantly with almost zero space usage;
- Pay-as-you-go: resources not in use are automatically released and bills are friendlier.
give an example::
When a retail company experiences a surge in traffic on Double 11 or Black Friday, Snowflake automatically scales the back-end compute cluster without lagging at all.
- How are storage and compute separated in Snowflake?
why did you take the test?: It depends on your understanding of its decoupled architecture to play to its advantage.
high score answer::- Storage is unified in the cloud (S3, Azure Blob, GCS, etc.), and computation executes queries through the Virtual Warehouse; the two scale independently of each other; in this way, performance can be increased as you go, and you can pay only for storage or computation.
typical example: CREATE OR REPLACE WAREHOUSE my_warehouse WITH WAREHOUSE_SIZE = 'MEDIUM' AUTO_SUSPEND = 300 AUTO_RESUME = TRUE;
- Storage is unified in the cloud (S3, Azure Blob, GCS, etc.), and computation executes queries through the Virtual Warehouse; the two scale independently of each other; in this way, performance can be increased as you go, and you can pay only for storage or computation.
- What is Snowflake's Time Travel?
why did you take the test?: See if you know how to use it to recover mistakenly deleted/historical queries.
high score answer::- Time Travel allows you to query and restore historical data within the retention period (up to 90 days); especially useful for auditing, debugging, and rollback.
typical example: SELECT * FROM sales_data AT(TIMESTAMP => DATEADD(HOUR, -24, CURRENT_TIMESTAMP()));
- Time Travel allows you to query and restore historical data within the retention period (up to 90 days); especially useful for auditing, debugging, and rollback.
- What is Snowflake's Data Sharing and why is it so great?
why did you take the test?: A core tool for real-time collaborative analytics across enterprises.
high score answer::- Allows sharing of data across organizations, in real time, without ETL or replication of data;
- Secure and efficient, the other party can check your data directly in their account.
application scenario: Healthcare organizations share desensitized patient analytics data, in real time, with the Institute, without the need for cumbersome data portability on both sides.
- What happened to Snowflake's Micro-Partitions?
why did you take the test?: Microsegmentation directly determines scanning efficiency and storage costs.
high score answer::- Snowflake automatically splits tables into small columnar files ranging from 50-500MB; only relevant partitions are scanned when querying, dramatically increasing speed.
typical example: SELECT SYSTEM$CLUSTERING_INFORMATION('sales_data');.
- Snowflake automatically splits tables into small columnar files ranging from 50-500MB; only relevant partitions are scanned when querying, dramatically increasing speed.
- How does Snowflake handle semi-structured data?
why did you take the test?: JSON, Avro, Parquet are very common, let's see if you can get a handful.
high score answer::- expense or outlay
VARIANTType storage of raw data, no need to pre-define complex schema; support for extracting fields directly in SQL through point operators or functions.
typical example: SELECT my_json_column:customer.name, my_json_column:customer.age FROM orders;
- expense or outlay
- How to optimize Snowflake query performance?
why did you take the test?: Performance tuning is the key to saving money and speed.
high score answer::- Rational Cluster Building (Clustering Keys);
- Utilizing results caching and metadata caching;
- Partition Pruning is achieved by trying to make the filtering conditions fall on micro-partitions.
- How does Snowflake do failover and disaster recovery?
why did you take the test?: Business continuity is a top priority.
high score answer::- Snowflake provides cross-region replication and automatic switching (Failover);
- Configure reasonable retention period (Data Retention) and recovery policy, material seconds recovery.
- Configure reasonable retention period (Data Retention) and recovery policy, material seconds recovery.
- What is Materialized Views and when is it used?
why did you take the test?: Commonly used scenarios that make hotspot queries faster, but with additional storage overhead.
high score answer::- Materialized views pre-calculate and store query results for fast subsequent reads; suitable for those scenarios with complex calculations such as aggregation, JOIN, etc. and high query frequency.
typical example: CREATE MATERIALIZED VIEW mv_sales AS SELECT customer_id, SUM(amount) AS total_spent FROM transactions GROUP BY customer_id;.
- Materialized views pre-calculate and store query results for fast subsequent reads; suitable for those scenarios with complex calculations such as aggregation, JOIN, etc. and high query frequency.
- How does Snowflake's Role-Based Access Control (RBAC) work?
why did you take the test?: Safety and compliance will always be a hard target for the big players.
high score answer::- Tie permissions to roles and then group users into roles for simpler centralized management; support granularity to database, table, and view level authorization.
typical example: GRANT SELECT ON TABLE customer_data TO ROLE analyst_role;.
- Tie permissions to roles and then group users into roles for simpler centralized management; support granularity to database, table, and view level authorization.
Snowflake Interview - Code High Frequency Questions
- Leetcode212 You come up here, you answer the questions, and then you're challenged on your solution, why you're using trie, and so on.
- The key value map has a lot of follow up
- A. Provide get set update, delete (directly implemented in python using dict)
- B. support ttl, if expire to delete (add a new property ttl) (key, value + ttl) or here use expir timestamp
- C. given a timestamp, how to find all valid key value pairs (for loop comparison)
- Given a board, two players take turns to play, and each player wins if their flags are four in a row. Ask if a player wins after playing at position (x,y) canPlayWin(board, x, y, player), not hard but needs to focus on simplicity of code.
- There are nnn (even) houses arranged in rows numbered 1,2,...,n. A painter has to color these houses with at most three colors and two conditions must be satisfied:
- Adjacent houses have different colors
That is, for any i, house i and house i+1i cannot have the same color. - Symmetrically located houses have different colors
A "symmetrical position" is a pair of houses at the same distance from each other - i.e., house number iii and house number n-i+1 must be different colors.
- Adjacent houses have different colors
If you're also preparing for a Snowflake VO -- we can help!
Programhelp The team specializes in technical post interview assistance and service coverage:
OA Ghostwriting & High Frequency Questions
Coding Wheel Voice Assistance + Follow-up Solution Preparation
System Design Real-time assisted prompting
HM Interview Program Packaging & Behavior Interview Framework Training
We've helped several students land at popular North American Fintech & Infra plants like Snowflake / Stripe / Robinhood / Databricks. If you've got a Snowflake interview on the way and don't want to do it alone--
Feel free to private message for question bank + assisted proxy interview opportunities, and arrange full process 1v1 support for you!
