PinnedAlberto López Serna·Sep 15LLM and Statistical Synthetic Data with Dataflow — Part 3: the common runtime — one worker pool, a…Self-hosted LLM generation on Apache Beam / Dataflow / BigQuery — Part 3 of the series behind the Apache Beam Summit 2025 session “Building…
PinnedAlberto López Serna·Sep 9LLM and Statistical Synthetic Data with Dataflow — Part 2: the type system, and how free-text is…Part 1 ended on one log line: the generation_plan, which says for every column of the target table which strategy will produce its values…
PinnedAlberto López Serna·Sep 1LLM and Statistical Synthetic Data with Dataflow — Part 1: why, what, and the shape of the machineSelf-hosted LLM generation on Apache Beam / Dataflow / BigQuery — the series behind the Apache Beam Summit 2025 session “Building Banking…
PinnedAlberto López Serna·Jun 10, 2024Drools ParDo and SCIO Dataflow: A Goodbye Microservices TaleA successful re-engineering in banking with a Drools ParDo using a KieContainer (JBoss) in Dataflow has been used to process buissness…
PinnedAlberto López Serna·Apr 24, 2024Avoid HTTP requests duplicates in Apache Beam with SCIO, a custom BaseAsyncDoFn and State and…StateBaseAsyncDoFn
Alberto López Serna·Sep 21LLM and Statistical Synthetic Data with Dataflow — Part 4: b1_rag — retrieval that seeds the prompt…Self-hosted LLM generation on Apache Beam / Dataflow / BigQuery — Part 4 of the series behind the Apache Beam Summit 2025 session “Building…
Alberto López Serna·May 10Part 1: F1 Telemetry and Tuning for your Spark cluster: the BigQuery Log Analytics setup that costs…I can do auto tuning (every day) for +10000 spark jobs and ~400 clusters. Seamlessly SRE, cost tracking/optimisations and serious balance…