Role-specific question banks
Curated drills for Data Engineers, Analytics Engineers, and AI/ML Engineers so you practice the patterns DE hiring teams actually use.
FREE
Get the 10 Hardest DE Interview Questions (with model answers)
The specific questions candidates struggle with most, answered the way senior engineers answer them. Free.
We respect your inbox and will never sell your email address.
What is inside the drill pack
Structured, repeatable interview reps built for candidates who want focused preparation, not content overload.
Answer frameworks and model responses
Clear response structures plus strong, average, and weak examples so you learn how to answer with precision.
Objective scoring rubrics
Interview-style rubrics for depth, clarity, and tradeoff thinking so you can self-review and improve fast.
SAMPLE DRILL
Here is exactly what you get
One of 60 drills. SQL, system design, Python, and behavioral questions.
SQL
Mid-level
20 min
Rolling 7-Day Revenue Average
The Question
We have a table called daily_revenue with columns: date (DATE), revenue (NUMERIC). One row per day, some days may be missing.
Write a query that returns each date with the 7-day rolling average revenue (current day + 6 prior days). Then explain: what changes if the average uses calendar days vs. recorded days only?
Model Answer
SELECT
date,
revenue,
AVG(revenue) OVER (
ORDER BY date
ROWS BETWEEN 6 PRECEDING AND CURRENT ROW
) AS rolling_7d_avg
FROM daily_revenue
ORDER BY date;Key insight: ROWS BETWEEN (not RANGE BETWEEN) is unambiguous with gap data. The window frame handles fewer-than-7-days automatically. The follow-up tests whether you understand calendar-day vs. recorded-day semantics and when each matters.
Scoring Rubric
Pass if:
- Uses ROWS BETWEEN correctly and can explain why (not just RANGE)
- Knows the window frame handles edge cases natively
- Distinguishes calendar-day vs. recorded-day averages with a real explanation
Red flags:
- Confuses ROWS and RANGE framing
- Solves with a self-join instead of a window function
- Adds CASE WHEN logic to handle edge cases the window handles automatically
SAMPLE
The full pack has 54 technical drills like this, plus 6 behavioral scenarios. See pricing below.
Module breakdown
Focused modules that mirror how modern data and AI interviews are run.
Modern Data Stack
dbt, Apache Iceberg, and DuckDB
15 questions
- How do you choose between dbt incremental strategies (append vs merge vs delete+insert)?
- What is a custom materialization in dbt and when would you build one?
- How do you write effective dbt tests beyond the built-in ones?
- Explain dbt sources vs staging models. Why separate them?
- What are dbt snapshots and when do you use them over regular models?
AI-Augmented Pipelines
LLMs, vector databases, feature stores
10 questions
- How would you use an LLM to automate data quality checks in a pipeline?
- What is structured output from an LLM and why does it matter for data engineering pipelines?
- When would you add a vector database to a data platform? What problem does it solve?
- What is a feature store and why does it matter for ML pipelines?
- How do you manage the cost of LLM API calls in a production data pipeline?
Advanced DE Topics (2026)
Kafka streaming, Iceberg, dbt unit tests, data contracts, feature stores
10 questions
- How would you design a Kafka pipeline that guarantees exactly-once delivery?
- Your Kafka consumer lag is spiking. What is your diagnosis and remediation playbook?
- You are migrating from Hive to Iceberg. How do schema and partitions evolve?
- What is a data contract and how do you implement one for streaming?
- Online vs offline feature stores and how to investigate feature drift.
Behavioral Scenarios
Stakeholder comms, incidents, data quality
6 scenarios
- A data pipeline fails 2 hours before an executive dashboard meeting. Walk me through your response.
- A business stakeholder insists a data number is wrong, but your pipeline shows it is correct. How do you handle it?
- You discover 6 months of historical data was silently corrupted due to a schema change. What do you do?
- You are asked to deliver a new data product in 2 weeks that realistically takes 6 weeks. What do you do?
- Two downstream teams have conflicting definitions of the same business metric. How do you resolve it?
Real-World Scenarios
Migrations, incidents, scale, and LLM QA
7 questions
- Your company is migrating from a third-party SaaS CRM to an internally managed system. How do you avoid breaking downstream analytics?
- You've inherited a 15-year-old Informatica ETL job. How do you rewrite it in dbt safely?
- Your primary ELT job failed silently 6 hours ago at 3 AM. Walk me through your response.
- Traffic spikes 10x during breaking news. How do you architect the reader-behavior pipeline?
- How do you validate LLM-based ticket classification and prevent drift over time?
AI Infrastructure
RAG pipelines, vector databases, LLM observability
12 questions
- How do you design a document chunking strategy for a RAG pipeline serving long-form PDFs?
- Walk me through the architecture of a production RAG pipeline, from ingestion to query response.
- When would you choose pgvector over Pinecone or Chroma? Walk me through the tradeoffs.
- Your vector search results are degrading over time. How do you diagnose and fix embedding drift?
- How do you measure and monitor hallucination rate in a production LLM application?
- Design the data infrastructure for a real-time LLM-powered recommendation system.
Why this works for DE and AI interviews
Most prep materials are generic. This pack tightens your practice to the exact skills that show up in data engineering and AI interviews.
Data Engineering track
- System design drills for pipelines, reliability, and data quality
- SQL and warehouse scenarios with performance tradeoffs
- Modern data stack: dbt, Apache Iceberg, DuckDB
- Behavioral scenarios for senior-level interviews
- Behavioral and product sense questions tailored to DE roles
AI and ML track
- Modeling questions with evaluation and error analysis
- AI pipeline engineering questions
- Experimentation and iteration drills for real production teams
- Communication prompts for stakeholder clarity and impact
RK
Built by a practitioner, not a content mill
Ryan Kirsch
Data Engineer at the Philadelphia Inquirer
I've conducted 30+ technical screens as a hiring lead and been through 12+ data engineering interviews across media and tech companies. The patterns repeat. The gaps are predictable. I built this drill pack because I couldn't find one that actually matched how these interviews work in practice.
Connect on LinkedInPick your prep level
Start preparing today. Access is instant after checkout.
Interview Drill Pack
$49
Instant PDF download
- Core drill pack with 54 technical and 6 behavioral questions
- Answer templates and model responses
- 7-day prep plan and mock interview script
🛡 7-day satisfaction guarantee. If it is not a fit, reply for a full refund.
FAQ
Answers to the most common questions.
Who is this for?
Data Engineers, Analytics Engineers, and AI/ML Engineers preparing for technical roles at data-driven companies. The drills target what senior DEs actually get asked: dbt incremental strategies, Apache Iceberg, DuckDB, Kafka streaming, data contracts, feature stores, LLM pipeline design, and system design for data infrastructure. Most useful if you are interviewing in the next 2 to 8 weeks.
Why pay $49 when I can find questions on LeetCode or DataLemur for free?
Free resources cover generic SQL puzzles. This pack covers what DE hiring panels actually test in 2025 and 2026: dbt incremental strategies, Apache Iceberg schema evolution, DuckDB, Kafka streaming, data contracts, feature stores, LLM pipeline engineering, and behavioral scenarios from real data orgs. LeetCode and DataLemur will not prepare you for "how do you handle schema evolution in Iceberg" or "walk me through your approach to testing an LLM-dependent pipeline." If you want to be ready for a modern DE role, not a generic SQL position, this is what you practice with.
Is this for beginners or experienced candidates?
Both. Beginners get structure and clear answer formats. Experienced candidates sharpen depth, speed, and communication under pressure.
How long does it take to complete?
The core plan is designed for 7 days. You can also use it as a reusable practice system for ongoing interview cycles.
Is this live coaching?
No. This is a self-serve drill pack with guided frameworks, model responses, and rubrics so you can practice independently.
How do I get access after purchase?
Immediately. You will receive a Gumroad download link in your confirmation email. Open it, download the PDF, and start drilling. No waiting, no batch releases.
What if it is not a fit for me?
If the pack does not match the scope described on this page, reply within 7 days of access and request a refund.
Ready to land your next DE role?
Secure checkout. Instant access after purchase.
You will receive a confirmation email and download link immediately.
Start drilling today. The 7-day prep plan begins the moment you open the pack.