Robin Moffatt Portfolio
Principal DevEx Engineer
Long-running blog (archive back to 2009) by Principal DevEx Engineer Robin Moffatt, covering Apache Kafka, Apache Flink, dbt, and DuckDB in hands-on technical depth, including practical integration write-ups like "Kafka to Iceberg — Exploring the Options."
Location
Somewhere on planet Earth, most likely near a laptop.
Skills
Notable
- Speaking at conferences since 2009, including QCon, Devoxx, Strata, Kafka Summit and Øredev
- Kafka Listeners - Explained (2018) is still listed among the blog's popular posts; the kafka-listeners repo has 118 GitHub stars
- Blog archive runs continuously from 2009 to 2026
What makes it work
A breakdown of the choices that make this portfolio stand out.
Integration-focused posts between specific tools
"Kafka to Iceberg" doesn't explain either technology in isolation — it addresses the actual integration question a data engineer building a streaming pipeline would have, which is a more useful framing than a generic tool overview.
A seventeen-year archive that tracks the modern data stack's evolution
Posts spanning 2009 to 2026 — OBIEE and Oracle performance still top the category list alongside Kafka Connect and ksqlDB — let a reader trace how the data-platform ecosystem actually changed over more than fifteen years, from someone writing about it in real time rather than retroactively.
Measured, tested takes on hyped tools
"Claude Code isn't going to replace data engineers (yet)" tests whether Claude Code can build a credible dbt project from real data, and a companion post builds an LLM-as-judge eval to score it. The verdict — worse than you if used wrong, worth adding to the toolbox if used right — comes from evidence, not reflex in either direction.
What Data Engineers can take from this
Specific, actionable tips to apply to your own portfolio — no generic advice.
Write about the specific integration between two tools your readers actually combine, not just a standalone overview of either one.
Keep a long-running, dated archive intact — seventeen years of posts let readers see how your thinking and the field evolved together, which is itself valuable context.
When writing about a hyped new tool, give an honest, specific assessment of what it does and doesn't do for your field rather than a purely promotional or dismissive take.
Case studies
The archive is the portfolio here: seventeen years of dated posts, a hit counter in the sidebar, and projects that mostly exist to be written up. Robin Moffatt applies the same method to Kafka listeners in 2018 and to grading Claude Code's dbt output in 2026: build it, break it, publish the notes.
Claude Code + dbt evaluation series
A connected run of posts: learning dbt with DuckDB ("Ten years late to the dbt party"), watching Claude Code build a dbt project, testing whether it can do so credibly from real data with and without dbt Labs' published skills and across Claude models, and building the eval to measure it ("Evaluating Claude's dbt Skills: Building an Eval from Scratch").
It pairs AI adoption with an evaluation method rather than anecdotes, the approach teams need before trusting agents with pipelines.
Compares with and without dbt Labs' skills
Crash course in LLM-as-judge built into the write-up
Projects
Kafka to Iceberg — Exploring the Options
A decision guide for getting data from Kafka into Iceberg. It compares three routes, Flink SQL, Kafka Connect (both open source) and Confluent Tableflow (paid), mentions others at the end, and flags Apache Spark as the conspicuous omission. It builds on earlier hands-on posts about writing to Iceberg on S3 with a Glue catalog from both Flink SQL and Kafka Connect.
The answer is "it depends", and the post spells out what it depends on. That is the reasoning a streaming-platform architect is hired for.
Open-source and paid options compared side by side
Earlier posts supply the hands-on setup for each route
Kafka Listeners - Explained
An explainer on Kafka's listeners and advertised.listeners configuration, one of the most common Kafka connectivity stumbling blocks. It comes with a companion repo, and a later Go post checks advertised.listeners programmatically.
Being able to explain the config that breaks everyone's first Kafka deployment is developer-experience work that saves support hours.
Still a popular post eight years after publication
Pinned repo to reproduce the scenarios
More Data Engineer portfolios
Ready to build your portfolio?
Follow the Data Engineer roadmap — skills, projects, and timeline to get hired.