Coinbase · Primly Community

how I'd prep for the Coinbase interview if I started over

qa_quinn · 5 replies

Just accepted an offer at Coinbase (data eng role, L5 equivalent). Took me about 7 weeks of active prep. Sharing what I'd actually do differently if I had to start from scratch today.

Quick background: 5 YOE in data science and analytics, applied for a data engineering role on the protocol data team. The loop was 5 rounds: a recruiter screen, a take-home SQL/analytics case, a technical deep-dive on the take-home, a data modeling / system design round, and a behavioral.

What I wasted time on:

I spent the first two weeks grinding Leetcode hard problems. For a data engineering role, this was mostly irrelevant. I hit maybe one medium-difficulty Python problem in the whole loop. If you're applying for an SWE role this might matter more, but for analytics/data eng positions the investment should be heavily weighted toward SQL, data modeling, and domain-specific design.

I also read a lot of generic interview prep content. Not useful.

What actually mattered:

SQL. Window functions, CTEs, query optimization, handling slowly-changing dimensions. The take-home case was a real data problem that required writing clean, well-structured SQL to answer a business question. No tricks, just genuine fluency. If your SQL is rusty, start there.

Blockchain data is specific. On-chain data has a different shape than typical transactional databases. UTXO model, address-to-address transactions, event logs from smart contracts. I spent two weeks just reading about how crypto transaction data is actually structured and what kinds of questions you can answer with it. This paid off significantly in the technical round.

Data modeling for financial systems. Understand slowly-changing dimensions, audit tables, event-sourcing patterns. Know why you'd pick one over another.

For behavioral: read Coinbase's cultural values document closely. I printed it out and mapped a specific story to each value. Seemed excessive but one of my interviewers explicitly mentioned one of those values by name during the round.

Timeline I'd use now:

Weeks 1-2: SQL fluency, data modeling fundamentals. Week 3: read everything I could about on-chain data structure, Coinbase's public engineering blog. Weeks 4-5: system design for data systems specifically, practice walkthroughs. Week 6: behavioral prep, map stories to values. Week 7: mock interviews, rest, tighten loose ends.

The take-home was the real filter. I've heard from others who got filtered there. Don't rush it, treat it like a real work product.

5 replies

de_derek

The take-home filter is real. I got cut there on my first attempt a year and a half ago. Was writing functional SQL but the structure was messy and I didn't include any commentary explaining my assumptions. Second time I treated it like I was writing something a coworker would review and got through.

ds_dmitri

exactly this. they're evaluating whether you communicate through your work, not just whether you get the right number at the end.

newgrad_neil

Is the take-home the same for new grad data roles or is that more of a senior candidate thing? I have a recruiter screen next week and trying to figure out what to expect.

apm_aisha

curious whether the PM loop has a similar take-home component or if it's all live case rounds. different role obviously but this is helpful for calibrating what "homework" means at Coinbase.

ml_mike

The engineering blog point is underrated. A lot of companies have public write-ups on what they actually built. Referencing those in a design round shows you did your homework without being creepy about it.