Announcing leetdata.ai — A Practice Ground for Data Engineers

Software engineers got a gym. Data engineers got production. It's time to fix that.

A software engineer preparing for a career walks a well-worn path: grind LeetCode, study system design primers, rehearse mock interviews on platforms built for exactly that purpose. An entire practice infrastructure stands ready — thousands of problems ranked by difficulty, instant feedback, a community sweating through the same exercises.

Now try preparing as a data engineer. You grind... the same LeetCode. The same array puzzles and binary tree inversions. Then you walk into an interview, and someone asks you to design a dimensional model for an order management system, or to reason through late-arriving data in a pipeline. Nothing you practiced shows up in the room. Nothing in the room showed up in your practice.

We built practice infrastructure for every software discipline except the one that moves the world’s data. Today, I’m doing something about it.

leetdata.ai is live. It’s a practice platform built for data engineers — real problems, data modeling challenges, mock design interviews, and a leaderboard to keep you honest. It’s open to everyone, starting now.

Why is Data Engineering Weekly launching it

For years, I’ve curated Data Engineering Weekly by reading what the industry actually ships — the architecture posts, the postmortems, the migration stories, the debates about modeling and orchestration. And for years, the same question has landed in my inbox from early-career readers: “This is great, but where do I practice?”

I never had a good answer. I could point them to blog posts about dimensional modeling, but reading about modeling is like reading about swimming. At some point, you have to get in the water, and there was no pool.

leetdata.ai is the pool. It’s not a prep site bolted onto a trend — it’s the answer to a gap I’ve watched widen with every issue of this newsletter.

Fundamentals matter more now, not less

Here’s the uncomfortable timing: I’m launching a fundamentals platform in the middle of an AI boom, when models write more of our pipeline code every week. Doesn’t AI make this obsolete?

The opposite. AI collapses the value of syntax recall and raises the value of judgment. When a model can generate a working pipeline in seconds, the engineer’s job shifts entirely to the questions the model can’t answer for you: Is this the right grain for the fact table? Should this be incremental or a full refresh? What happens when the upstream schema changes? What does “correct” even mean for this dataset?

Engineers who understand systems from first principles direct AI. Everyone else takes direction from it. Engineers entering this field today will spend their careers reviewing, correcting, and architecting around machine-generated code — and you cannot review what you cannot reason about. Fundamentals are the leverage.

Practice is the job

Here’s what makes leetdata.ai different from generic interview prep: the practice and the work are the same thing. Partitioning strategies, dimensional modeling, pipeline design, data quality tradeoffs — these aren’t proxies for the job. They are the job.

When you solve a data modeling problem on leetdata.ai, you’re rehearsing the exact conversation you’ll have in a design review six months into your first role. Every problem you complete makes you better at the interview and better at the work waiting on the other side of it. That’s the whole point.

What you’ll find inside

Data engineering problems. SQL, transformations, pipeline logic — the day-to-day craft. Solve each problem in SQL, Python, or PySpark, filter by topic — joins, aggregation, window functions — and work through the same questions companies like Databricks, Netflix, and Stripe ask in their interview loops. Each problem answers the question every engineer quietly asks: can I actually do this work?

Data modeling problems. Modeling is the most under-trained, most interview-critical skill in data engineering. It’s the round that filters out most candidates, and until now there was nowhere to practice it deliberately. On leetdata, you build the schema yourself on an interactive canvas — add tables, draw relationships, define cardinality — against real scenarios like designing a denormalized search index for an e-commerce catalog. Then you validate your design and compare it against a reference. This is the heart of the platform.

Mock design interviews. Knowing the answer and holding a design conversation under pressure are different skills. The mock interviews let you rehearse the second one — a timed session, a full problem context, and an AI interviewer you can talk through your approach with, exactly like the real round. Tradeoffs, follow-up questions, thinking out loud — practice it before it counts.

A progress dashboard. Not streaks and vanity metrics. The dashboard shows which capabilities you’re building and where the gaps are, so your practice compounds instead of wandering.

A leaderboard. See where you stand among peers solving the same problems. A little competition sharpens practice — and the people beside you on that board are your future colleagues.

A daily problem. Every day, one problem. No decision fatigue about what to practice next — open the daily, solve it, and let the habit compound. You don’t build fundamentals in a weekend binge; you build them in twenty focused minutes a day.

Two asks

If you’re building your data engineering career: sign up and open your first problem. Start with the supply chain shipment tracking data model. It looks simple until the shipment updates its status out of order, and then you learn firsthand why event ordering shapes every modeling decision you’ll ever make. That moment — where the problem teaches you something the tutorial never mentioned — is what leetdata.ai delivers, problem after problem.

If you’re a senior engineer reading this: this platform is for you too. Fundamentals rust quietly — when did you last design a schema from a blank canvas instead of inheriting one? Work through a modeling problem and see if your instincts still hold, or run a mock interview before your next job move. And if the problems remind you of someone on your team who’d benefit, send it their way.

Back to first principles

Data Engineering Weekly has always been about understanding this field deeply rather than skimming it. leetdata.ai extends that mission from reading to doing. The industry left data engineers without a practice ground for too long, and the timing has never mattered more — the engineers who master fundamentals now will be the ones directing the AI-augmented data teams of the next decade.

The platform is live. Your first problem is waiting.

Start solving at leetdata.ai.