August 25 Update

TrustEdge now supports a robust envelope file format (.trst), per-chunk Ed25519 signatures, and authenticated metadata (AAD) for streaming workloads. The CLI now supports flexible encryption, decryption, and key management options. Each chunk is cryptographically bound to its metadata and provenance. Feedback is welcome, and a future post will dive deep into the envelope and manifest design!

Major Update

TrustEdge now supports a real encrypted envelope file format (.trst) with per-chunk Ed25519 signatures, robust AAD, and flexible CLI options for encryption, decryption, and key management. Each chunk is cryptographically bound to its metadata and provenance. A follow-up post will dive deep into the envelope and manifest design!

Day 2 Update

> TrustEdge now features improved nonce discipline, authenticated headers (AAD), tamper checks, and key zeroization. Lessons learned include the importance of unique nonces, binding metadata, and careful header handling. Stay tuned for a full “Day 2” deep dive post!

title: TrustEdge — Trustable Edge AI (Rust) date: 2025-08-19 tags: - Rust - Edge AI - Privacy - Encryption - IoT - Project


TrustEdge — Trustable Edge AI (Rust)

Announcing my latest project: TrustEdge! This is my public learning journey in Rust, focused on building privacy-preserving, trustable edge AI pipelines. Instead of a CRUD web app, I’m exploring IoT, security/PKI, and edge systems, with a strong emphasis on privacy by design—encrypting at the edge, not just in transit.

What’s in Phase 1?

  • Minimal CLI demo: chunked file read, per-chunk AES-256-GCM encryption/decryption, immediate verification
  • Written in Rust for safety and performance
  • Honest, incremental milestones — real, reviewable code

Why TrustEdge?

If you’re into Rust, IoT, ML at the edge, or security, I’d love your feedback! The code is open and the journey is public.

Check out the TrustEdge repo on GitHub