sanctions-compliance

Deterministic evidence-packet linter for claim-backed AI output. Reports packet completeness, not truth. CLI, Python API, MCP, A2A, Cloudflare Workers.

A Claude Skill that adjudicates sanctions / PEP / adverse-media screening hits deterministically. Token-efficient tiered design (parse → cheap FP rules → identifier corroboration → targeted multilingual research). Handles 9+ naming conventions, transliterations, and historical place names.

AI Swarm system for petroleum & petrochemical trading — 6 specialized agents for market analysis, logistics, compliance, documentation, finance, and operations management

On-chain forensic analysis & FinCEN Suspicious Activity Report (SAR) narrative suite investigating the $625M Lazarus Group / Ronin Bridge exploit.

End-to-end Sanctions Screening & Compliance Analytics project using SQL, Python, Tableau, and Streamlit to analyze watchlist matches, risk scoring, false positives, escalations, and sanctions exposure.