Open source engineer building data platforms, developer tooling, and technical writing around Rust, Python, Go, and modern analytics systems.
Website | Blog | LinkedIn | Email
Systems I actively build. Related papers, integrations, datasets, and distribution repositories are grouped with their parent project.
| Project | Focus |
|---|---|
| dataprof | v0.13.1: bounded-memory Rust/Python profiling and pipeline quality gates for CSV, JSON, Parquet, DataFrames, and Arrow streams, with explicit assessment scope and comparable quality reports. Related work includes Metric Evidence, a Power BI report with evidence beside each metric; a dlt/SQL/DuckDB pipeline reference; a seven-rule, six-tool comparison of checks and costs; and an ADBC query profiling and type/stream matrix. |
| arrow-abi-doctor | Differential testing of the Arrow C Data and C Stream interfaces across real consumers, in isolated processes. |
| Tabular Evolution Corpus | v0.3.0: versioned Parquet scenarios of datasets changing shape, with machine-checked manifests, also on Hugging Face. |
| Nephtys | A low-footprint, multi-protocol connector for durable NATS JetStream pipelines, backed by an IEEE UIC 2026 evaluation. |
More selected work: open data, knowledge graphs, lakehouse, applied research, and systems work
| Project | Focus |
|---|---|
| Ceres + Ares | v0.7.0 / v0.4.0: reproducible open-data harvesting and schema-driven extraction. Ceres powers a public open-data index and a Databricks analytics pipeline. |
| GitNodes | v0.2.0: a Git-native, typed knowledge graph for engineering teams and AI agents, accessible through the web UI or MCP. |
| dbt-isrm | Active: a release matrix for dbt's Information Schema across releases and execution stages. |
| dlt + dbt on Databricks | Active: runnable ingestion, transformation, orchestration, and quality-gate examples validated against Databricks and in DuckDB CI. |
| OCCAS | v0.1.0 research release: a provenance-backed PX4/ERA5 agreement study; external wind estimates were too imprecise to substitute for onboard estimates under the study's criterion. |
| Fantabuddy | v0.4.0 data product: a local DuckDB history of official player lists, offline Classic/Mantra auction reports, and verifiable Parquet corpus exports. |
| DCE | Pre-release: a Rust-native engine for defining, validating, and enforcing data contracts. |
| Zero Grappler | Prototype: a no_std, allocation-free async pipeline for embedded sensor, DSP, and inference workloads. |
| IcebergSharp | In development: .NET Iceberg metadata, Avro manifests, and REST catalog access; scan planning and Parquet data reads remain on the roadmap. |
| Lares | In use: a small self-hosted home stack built around failure semantics, with digest-pinned images, application-consistent SQLite snapshots to encrypted offsite backup, and a scripted bare-metal rebuild. |
Browse the full project archive or all repositories.
- 2026-11-09: One Year, 24 Posts: What Held Up and What Didn't
- 2026-10-20: Where Should the Library Stop?
- 2026-10-05: Green Is a Policy, Not a Fact
The full blog has posts in English and Italian.
Contributions to the broader open source ecosystem beyond my own repositories.
This section is updated automatically via GitHub Actions.
64 merged PRs across 29 upstream projects · 221.9k combined stars
