Mission briefing
CodeKitchen is a free, online, individual data-engineering hiring contest run by Quera together with DataChef. You get 6 problems in 2h 30m. Strong performers are invited to interview for a remote "Chef" (data engineer) role.
What the official page says
Problems
"Data engineering challenges": you analyse the problem, choose an approach, and implement it.
Minutes
That's about 25 minutes per problem on average, so speed and accuracy both count.
Prerequisites
"Familiarity with Python and database concepts." It's open to both working data engineers and people still learning.
Outcome
Top people get an interview. If both sides agree, you work remotely from where you live, with conditions based on your level.
Who is DataChef?
DataChef is a boutique Data & AI consultancy. Former Amazon managers founded it in 2019 in Amsterdam (it now lists an office in The Hague). It has 40+ engineers, called "Chefs", and a large Iranian team: their GitHub even hosts a Farsi relocation-to-netherlands guide. Its job is to "put AI into production and make it pay off". In practice that means building data platforms that clients can run and maintain themselves.
🧰 Tech stack they advertise
Languages: Python 3, Scala, TypeScript (AWS CDK)
Processing: Apache Spark (they wrote sparkle, a Spark meta-framework), Kafka, dbt, Airflow
Cloud: AWS (S3, Glue, Lambda, Lake Formation, Athena, Kinesis), GCP
Platforms: Databricks, Snowflake, Fivetran, Datadog, Terraform, Kubernetes
🏗️ What they build
Data Mesh & Data Lake / Lakehouse platforms on AWS (their aws-data-landing-zone CDK construct)
Platform modernization, IoT data, predictive analytics, AI in production
Clients: Porsche, VodafoneZiggo, Rituals, PostNL, FrieslandCampina, Syngenta, Brenntag, ANWB…
Culture: "senior-only", "same-side, no-slides", owner mentality.
What this means for the contest. They are hiring data engineers, not ML researchers or competitive programmers. Expect tasks like those in a DE job: write correct SQL (joins, aggregations, window functions, dedup, sessionization), transform messy data in Python (parse, clean, validate, aggregate, flatten JSON, logs, CSV), and possibly pipeline logic (idempotent loads, SCD, DAG ordering) or conceptual questions about databases and data platforms.
Likely hiring process
Only the contest and the take-home come from public sources. The middle steps are the usual consultancy pattern.
Past DataChef job posts list AWS (Lambda, Glue, Lake Formation), Spark (Python/Scala), Kafka and "data management with big data". They want an all-round consultant, not just a cloud specialist. Lessons 08 and 09 cover enough of these concepts for the contest and the first interview.
Have there been similar contests?
This appears to be the first CodeKitchen. I found no earlier editions or published problems. The closest precedents are Quera's own data contests, such as the CodeCup 6/7 data track. That track had separate contests for Database (MySQL queries), Software engineering (plain implementation in any language), Data analysis and ML, each with 3–7 problems from easy to hard and automatic judging. Treat that as the template, minus the heavy ML, plus pipeline-style problems.
Your plan for the next ~3.5 hours
| Block | Lessons | Why |
|---|---|---|
| ⏱️ 0:00–0:20 | 00 Briefing, 01 Format & strategy | Know the judge and avoid silly point losses |
| 🐍 0:20–1:10 | 02 Python toolkit, 03 Wrangling, 04 pandas | Most problems will be solved in Python |
| 🗃️ 1:10–2:00 | 05 SQL core, 06 SQL advanced, 07 DB concepts | "Database concepts" is explicitly required |
| 🏭 2:00–2:25 | 08 DE concepts, 09 Algorithms | For pipeline-style and conceptual questions |
| 🧪 2:25–3:15 | 10 SQL lab, 11 Python lab | Live-judged practice, the most valuable part |
| 🏁 optional | 12 Mock contest | Six new problems against a timer, if you have an extra hour |
| 📋 3:15–3:30 | 13 Cheat sheet | Then eat, drink water, and rest ☕ |
Short on time? Follow the critical path: 01 → 06 (SQL windows) → 03 (wrangling) → 10 & 11 labs → 13. Skim the rest.
Only about an hour left? Read the callouts and tables only (they hold the traps), do the medium exercises in 10 and 11, and finish with 13. Skip the concept lessons: they matter more for the interview than for the contest.