On this page
Most guides about getting into a big company are written for software engineers. This one is for a data engineering role: the person who moves data from where it is created to where people can use it. It explains what Google interviews usually look like, what to practise, and how to spend 8 weeks. You do not need to know any of this already.
One page, with one or two projects you can explain.
A short chat about you, the role and the timeline.
Coding and SQL, solved while you talk.
Several interviews in one day or a few days.
Feedback is reviewed together before an offer.
What the role looks like
At Google, data engineers build and run the pipelines and tables that other teams rely on for reports, products and experiments. Job titles and team names differ, so read each posting for its own list of skills instead of assuming one standard job.
Google also runs internship programmes for students, and these are a common first step for freshers. Check the careers page for the programmes open to your year and country.
How the hiring process usually runs
Read this as a typical order, not a promise. Some stages are skipped for some roles, and for students the first stage is often an internship application.
- Application and resume screen. A person or a tool checks that your skills match the posting. Keep the resume to one page and name the tools you used (SQL, Python, a database).
- Recruiter call. A short call about your background and interests. Have a 30-second answer for "tell me about yourself" ready.
- Technical phone or video screens. You write code or SQL while talking. Expect easy to medium problems. Speak your reasoning out loud.
- The interview loop. Several interviews, often a mix of coding, SQL or data questions, design thinking for data, and questions about how you work with others.
- Review. Interviewers write feedback and a group reviews it. You may be asked to talk to teams about where you would fit.
What they test and how to practise it
Each row is one thing an interviewer wants to see, what to practise for it, and where to practise it here.
| What they check | What to practise | Practise it |
|---|---|---|
| Coding fundamentals | Hash maps, sorting, two pointers, simple recursion | Python practice |
| SQL | Joins, GROUP BY, window functions, finding duplicates | SQL practice |
| Data thinking | What one row means, how to check a result | Learn: SQL track |
| Pipelines | Raw, clean and reporting layers, reruns that are safe | What is a data pipeline? |
| Working with people | A short story about a problem you solved in a team | Write 3 stories and practise them aloud |
Tools and ideas worth knowing by name
Google describes several of its data products publicly. Knowing what they are for helps you follow the conversation.
- BigQuery is a data warehouse you query with SQL. Practise standard SQL and the idea of reading only the columns you need.
- Dataflow runs data processing jobs on the Apache Beam model, for both batch and streaming data.
- Pub/Sub is a message service that carries events from one system to another, like Kafka.
You do not need to be an expert in any of these. You need to say what each one is for, in one sentence, and show that you can learn it.
Questions to practise
These are the shapes of question to practise, written the way an interviewer might say them. They are not a leaked list. Solve each one yourself first, then check your answer.
- Given an orders table, return the top 3 products by revenue in each category.
- Find the customers who placed an order every month this year.
- A table has duplicate rows for the same event. Keep only the latest one.
- Count how many times each word appears in a list, and return the most common three.
- Merge two sorted lists into one sorted list.
- Your daily table is missing yesterday's data. How do you find out, and how do you fix it safely?
How to show clear thinking
Google-style interviews often reward the way you break a problem down. Restate the problem, give a small example, say the simple way first, then improve it. Say the time cost of your solution in plain words, such as "this loops through the list once, so it grows in step with the list".
An 8-week plan
Study about an hour a day. If you have less time, stretch each row and keep the order.
| Weeks | Focus | What you can do at the end |
|---|---|---|
| 1 to 2 | SQL basics: joins, GROUP BY, NULL | Answer easy SQL questions without looking anything up |
| 3 to 4 | Window functions, CTEs, deduplication | Solve "top N per group" and "latest row per key" questions |
| 5 | Python: dictionaries, loops, strings, files | Solve easy coding questions in about 20 minutes |
| 6 | Pipelines and warehouse ideas | Explain raw, clean and reporting layers and how a rerun stays safe |
| 7 | One project, written up | Walk through the project in 5 minutes and answer "what went wrong?" |
| 8 | Mock interviews for Google | Talk through a problem out loud while you solve it |
Mistakes freshers make
- Memorising answers. Interviewers change the data and the wording. Learn why a solution works, not what it looks like.
- Going silent. Say what you are thinking. A wrong answer with clear thinking often scores better than a right answer nobody understood.
- Skipping the checks. After you write a query, say how you would check it: count the rows, look for NULLs, look for duplicates.
- Having no project. One finished project you can explain beats five half-finished ones.
- Not asking questions. If a question is unclear, ask about the data and what one row means. That is what a working data engineer does.
How to talk about your project
Pick one project and prepare a 5-minute story: the problem, the data, the design, one thing that went wrong, and what you would change. Interviewers at any company, including Google, tend to spend the most time on "what went wrong", because it shows how you think. Build one from the projects page and write it up with a README.
Where to start today
Open the SQL practice problems, solve three, and write down what confused you. Then pick the Learn track that covers it. A small start today is worth more than a perfect plan next month.