Case studies
Proof of Work
A selection of data systems built for repeatable, reliable analysis.
Client work
Regulated market data · 300+ production pipelines
Regulated market-data ingestion platform
Cloud ingestion and transformation behind 300+ production pipelines, spanning regulated-business records, geospatial sources, APIs, and PDF documents.
Runs as the governed ingestion standard behind 300+ production pipelines — auditable, reproducible, and no longer dependent on one-off scripts.
Public health · 2M+ people reached
Public-health risk data services
Statistical components, data pipelines, and serverless backend services behind a public COVID-19 mortality-risk application used by 2M+ people.
Backs a public risk application serving 2M+ people, with serverless services that hold up under real public-load spikes.
Insurance · Daily and monthly incremental snapshots
Insurance analytics data platform
Tested warehouse models, compliance datasets, and analytical marts on a fixed daily and monthly cadence, replacing manual extracts behind regulatory filings.
Gives compliance and actuarial teams reconciled, trusted marts on a fixed daily and monthly cadence instead of manual extracts.
Veterinary healthcare · 46 clinic locations
Multi-location veterinary inventory ETL
A modular ETL package that validates inventory adjustments and mappings across 46 clinic locations, on reproducible weekly releases.
Automated validated inventory adjustments and reproducible weekly releases across all covered locations.
LegalTech / Regulated eDiscovery · $1M+ engagements
Regulated eDiscovery data ingestion
Ingestion, validation, and records reconciliation for litigation and investigation datasets that had to hold up as evidence.
Large litigation and investigation datasets that held up downstream and in discovery, validated and reconciled before review.
Independent work
Maintained projects
Setu
A Rust data-activation engine that reacts to PostgreSQL change streams and delivers matching events to webhooks, Slack, or Telegram without polling or middleware.
Rental Market Dynamics — Dubai
An automated pipeline for extracting Dubai rent-contract data, transforming it to Parquet, and publishing analysis-ready releases and property-usage reports.
Hermes Google Sheets
A maintained Hermes Agent plugin for reading, searching, and updating Google Sheets through focused spreadsheet tools.
Contributor work
Open-source contributions
PUDL
Submitted upstream work to modernize six test modules from unittest patterns to pytest, move FERC pipelines from SourceAsset toward AssetSpec, and iterate on nightly-build cache handling in PR #2951 and PR #2953.
sportsdataverse-py
Submitted an NFL play-by-play analysis example covering field goals, kickoffs, and punts.
These submissions were reviewed upstream and closed without merge.