/ PROJECTS
Tools, datasets, and research at the law × AI seam
A working portfolio — legal-AI products, privacy models, benchmarks, and the open-source infrastructure underneath them. Most ship with code, demos, or an essay.
/ FEATURED
Legal AI · BetaCicero
A legal workbench that turns messy inputs — email threads, third-party paper, playbooks, quick requests — into redlines, issue lists, memos, and matter context.
Open project →
Benchmark · 2026benchmarks.law
An independent legal-AI benchmark: tools tested without vendor consent, results never withdrawn, no LLM judges. The method is published; first results are pending.
Open project →6,427Word documents scored per converterBenchmark · 2026neurotic_docx_bench
An open benchmark that scores DOCX converters and redline engines against Microsoft Word's own output — 6,427 documents, 3,502 compare pairs, failures count as zero.
Open project →
Privacy · 2026PII Detection Models (pt-BR / LGPD)
Open models for finding personal and LGPD-sensitive data in Brazilian Portuguese text, with a live demo and a benchmark of recall against false positives.
Open project →56.7%of lawyers' first drafts rated reliable. The 13 AI tools averaged 57%.Research · 2025Benchmarking Humans & AI in Contract Drafting
Co-authored study scoring 13 AI tools and in-house lawyers on 30 contract-drafting tasks: 450 outputs, 72 survey responses, 12 interviews.
Open project →/ ALL PROJECTS
Brazilian Realistic Data Generator
Realistic synthetic pt-BR PII — names, addresses, and documents — for privacy testing and model training.
Open project →Dataset · 2026Material Contracts Corpus
Material-contract (EX-10) exhibits from SEC filings, published as a dataset.
Open project →Field notesThe Synthetic Lawyer
Essays on legal AI, contracts, privacy, and making legal systems work.
Open project →Essay · 2024Taming 12,773 Contracts
Large-scale LLM extraction over material-contract exhibits — for $39.26.
Open project →