Skip to content
View logickeau's full-sized avatar

Block or report logickeau

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
logickeau/README.md

Jean "Nick" Pijeau

AI Security & Integrity Engineer
I build Vitnify, a deterministic LLM engine (bit-identical across CPUs and GPUs) and tamper-evident, offline-verifiable execution receipts for AI agents.

vitnify.com · LinkedIn · Orlando, FL


🔭 What I'm building

Vitnify is my answer to the question I spent years asking as a threat hunter: what actually happened? It lets you re-run an AI agent's model and get bit-identical output, then check a signed record of the whole run offline. Built end to end:

  • 🧠 vitni-tensor: a deterministic, no_std Rust inference engine with bit-identical LLM output across CPU vendors, instruction sets, and GPUs. Verified byte-for-byte on Apple, AWS Graviton, Intel, AMD, Metal, and CUDA (pinned-order reductions + exact-integer k-quant kernels).
  • 📜 vitnify + vitnify-receipt-spec: an open protocol and SDK for tamper-evident, offline-verifiable agent execution receipts (BLAKE3 Merkle + ed25519), with LangGraph and MCP drop-ins. pip install vitnify.
  • ✅ conformance: a language-agnostic corpus so any implementation can be held to the spec.
  • 🔐 A Verification Authority service + multi-tenant platform behind it (FastAPI · ed25519 · Postgres).
pip install vitnify        # Python SDK
cargo add vitni-tensor     # the deterministic engine

🧭 How I got here

I've been hunting threats my whole career, just in different domains. As a combat engineer, I cleared mines and IEDs off the roads and hunted the enemy on recon patrols. As an Air Defense Artillery officer, I tracked incoming aircraft and missiles. Cyber ran alongside all of it: app security, CTFs, and constant training while in uniform, then full-time threat hunting and incident response (CISSP). Same instinct every time: find the thing that shouldn't be there. Now I build the tools that make an AI's run something you can verify, not just trust.

🛠️ Stack

Rust · Python · TypeScript · LLM / GGUF / quantization · cryptography (ed25519, BLAKE3, Merkle) · AI-agent & MCP security · deterministic / reproducible ML · AWS · PostgreSQL

I'd rather build it than diagram it.

Popular repositories Loading

  1. navicourse navicourse Public

    MGRS - Land Navigation Course Generating tool

    PHP

  2. Factoric-Permutation Factoric-Permutation Public

    A CS problem asked to be solved during a job interview

    C#

  3. Disk-Based-Data-Logger Disk-Based-Data-Logger Public

    Assembly

  4. Quad Quad Public

    the Quadzer APP in the making!

    Java

  5. MilClock MilClock Public

    An Open Source Military Time Synchronization widget

    Java

  6. logickeau logickeau Public

    Profile README