👋 Welcome to my technical blog!
I write about Data Engineering, Rust, Go, Python and open source technologies.
Exploring lakehouse architectures, real-time streaming and the modern data world.
dbt rewrote itself in Rust. The speed is the boring part.
I opened a bug report against dbt and found a codebase nobody had told me about: 77 Rust crates where the Python used to be. This is what the rewrite is actually for, why the marketing undersells it, and what I broke loose while poking at it — two fixes, both for bugs that never printed an error.
Nephtys on a Raspberry Pi: 6.6× less memory, exactly the same watts
I had never publicly written about Nephtys, and the first thing I have to report about it on real hardware is a null result: the memory advantage reproduced on a Raspberry Pi 5, the energy advantage never existed.
Zerobus Ingest: when Unity Catalog becomes the sink
In the dlt+dbt post the contract was a Unity Catalog schema. With Zerobus Ingest I moved the same question to the producer edge: when the sink is already a governed Delta table, do you really need a bus in the middle?
dlt + dbt on One Unity Catalog (and Why It's Not 'Two DLTs')
Python-native ingestion with dlt (dlthub), SQL transformation with dbt, and a single Unity Catalog schema as the contract between them. Notes from wiring them together on a real Databricks workspace — including the naming collision that bit me in actual code.
Brain and Brain UI: A Knowledge Base You Actually Own
Brain is a Git-backed Markdown knowledge base. Brain UI is the Rust/Leptos control plane on top of it. This is the system, the connection between the two halves, a real config-driven taxonomy, and the honest reasons I’m putting the core in the open.
Building My Personal Mini-Site as a Real Project
How I rebuilt my personal site as a versioned, reproducible static-site system with clear separation between landing page, blog, generators, and a small Go API companion.
FinOps for AI Data Platforms in 2026: Databricks vs AWS-Native vs DIY Iceberg on Top of Your Warehouse
Three lakehouse stacks for adding AI/ML and streaming on top of an existing warehouse, compared through a FinOps lens: DBUs, DPUs, TB-scanned, S3 GB-month, and egress, walked through real workload patterns with 2026 prices.
Zero Grappler: Data-Pipeline Thinking on a Microcontroller (Draft Notes from Before the Hardware Arrives)
Zero Grappler is a small no_std crate that applies a data-pipeline mindset to embedded ML: three traits, two async tasks, compile-time buffer sizing, zero allocations. This post is about the design choices — not yet a hardware report. The Pico 2 W smoke test on real silicon is still ahead of me.
Lance Format and LanceDB: Columnar Storage for the Embedding Age
Lance is a columnar storage format built for machine learning workloads — fast random access, native vector indexing, and zero-copy Arrow integration. This article walks through the format itself, how LanceDB builds on top of it, and how I wired it into a live NATS stream to build a simple semantic search layer over real-time events.
Guardrails for Tabular ML: A Data Engineer's Take on Data Leakage, Poisoning, and Brittle Pipelines
Most ML pipeline failures are not exotic model bugs — they are data issues that nobody encoded as checks. This article walks through building guardrails using pandas, Apache DataFusion, data contracts, and the Arrow C Data Interface.