Enterprise security · Jun-Jul 2026, ongoing
Shadow-AI discovery & DLP platform
Multi-tenant, on-premise platform that discovers, classifies and controls AI-tool usage across managed devices. Built as a monorepo spanning five languages and runtimes.
What it did, and what it measured
- Six-service FastAPI backend: a latency-critical policy verdict engine, telemetry ingestion batched through Redis Streams, a signed tool catalogue, tenant/device control, DLP and a stream processor.
- Cross-platform Rust desktop agent (Cargo workspace, local SQLite, rustls-only TLS) plus a Chrome MV3 extension built with deliberately minimal permissions for store review.
- Detection designed around DNS, SNI and process events with no TLS interception - an explicit privacy boundary, documented in ADRs and the scope contract rather than left implicit.
Stack
RustPython 3.12 + FastAPITypeScriptNext.js 15 / React 19PostgreSQLRedis StreamsClickHouseAuth0TerraformTurborepoSemgrepTrivy
Repo and demo
This is client work, so the source is not public. Happy to walk through the architecture, the scorecards and the trade-offs on a call.