AlexP | Data Engineer

A new cohort of LLM Zoomcamp starts on June 8, 2026. It’s a free 10-week course where you go from LLM basics to building a production-ready AI assistant. For this cohort, I'll update the course content during a series of live workshops. In the course, you'll learn: - Retrieval-Augmented Generation - Vector search and embeddings - AI agents - Function calling and tool use - Evaluation of RAG and agentic systems - Monitoring LLM applications Join the new cohort and build your LLM application step by step: github.com/DataTalksClub/…

Add June 8, 17:00 CET to your calendars! github.com/DataTalksClub/…

What I like most about this setup: reproducible analytics versioned transformations tested data models static public delivery low-friction deployment It’s a good pattern for shipping public data products from an engineering-first stack.

Tech stack: • Prefect orchestrates the pipeline • ClickHouse stores and serves analytics fast • dbt handles transformations and testing • Evidence turns curated data into a static analytics site • GitHub Actions / Pages automate delivery

Built an end-to-end analytics project for #DEZoomCamp It turns raw GitHub repository metadata into a dashboard using: Prefect for orchestration ClickHouse as the warehouse dbt for tested transformations Evidence for the dashboard Pages for publication github.com/aipavlo/github…

Got this message from a Data Engineering Zoomcamp participant. John-Luke landed their first junior DE role while still working on the final project. Always great to see participants finding real value in the material!

Module 7 is a solid step into real-time data engineering: not just moving events, but actually extracting insight from live streams. Definitely one of the most engaging modules in the Zoomcamp so far

The most interesting part was working with tumbling and session windows. That’s where streaming starts to feel different from classic batch analytics

Just wrapped up Module 7 of DE Zoomcamp - a really hands-on introduction to streaming data pipelines with Kafka-compatible tools, PyFlink, windowed aggregations and real event-driven processing. #flink #kafka #dataengineering

Spark feels heavy until you start thinking in partitions + transformations + shuffle + actions After this homework, I’m more confident in reading/writing Parquet and debugging jobs via Spark UI

Zone lookup join: Load taxi_zone_lookup.csv into Spark, join on location ID, then find the least frequent pickup zone. This is a clean join + groupBy/count exercise and shows how to enrich raw facts with dimensional data

#DEZoomcamp (2026) Module 6 - Batch processing with Apache Spark. This homework was a great “Spark fundamentals in practice” run: Parquet, partitions, DataFrames, Spark UI, and some real NYC taxi analytics

@DataTalksClub @dltHub Quick sanity checks via SQL: - date range from trip_pickup_date_time - % of credit card payments from payment_type - total tips from tip_amt If you’re learning ingestion/ELT, this workshop is a great win for you. Course repo: github.com/DataTalksClub/… #DEZoomCamp

@DataTalksClub @dltHub Another useful detail: some columns received no data during a load, so dlt couldn’t infer types. Solution: add type hints in dlt.resource(columns). After that, everything materializes cleanly and is easy to query in DuckDB. #dataengineering

Just completed the @DataTalksClub Data Engineering Zoomcamp dlt workshop with @dltHub I built a small ingestion pipeline from a paginated REST API to DuckDB using dlt. Thread with the key steps and some insights