Skip to content
View MoAz06's full-sized avatar

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull SharkAchievement: YOLO

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull SharkAchievement: YOLO

Block or report MoAz06

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. 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
MoAz06/README.md

› Mohamed Azahrioui

Backend & AI-security developer — I build deterministic, auditable guardrails for AI agents.

The idea behind everything I build: let a deterministic rule engine make the decision, and let the model only explain it. The moment the model gets to decide, you've traded a verifiable answer for a confident guess — and in security and agent actions, that trade isn't worth it. So every decision ships as proof you can verify yourself.

backend developer · The Hague, NL · CS @ Leiden · shipping production code at CodeHive + Kojac

Selected work

ReachGate — vulnerability-reachability triage on GitLab Orbit

Walks the code graph with a bounded BFS to answer what scanners can't: can this vulnerability actually be reached from the app's entry points? On the codebases I tested ~80% of "critical" findings came back not reachable. Deterministic verdicts, UNKNOWN when unproven, OpenVEX/SARIF exports, an sha256 manifest + optional Ed25519 signature, and an offline verifier that needs no token and no network.

Python · BFS · OpenVEX · SARIF · Ed25519 · MCP · pytest

TrustGate — runtime authorization gate for AI agents

Sits between an agent and a risky action (like a refund) and checks live business evidence before it runs: ALLOW, APPROVAL_REQUIRED, or BLOCK, each with an auditable receipt. The model proposes; the policy engine decides.

Node.js · Google Cloud Run · Vertex AI / Gemini · BigQuery · Fivetran · MCP · React

Writing

I write about agent governance, reachability, and data-supply-chain trust — the questions that show up once agents meet real operational data.

Stack

Languages Python · TypeScript · JavaScript/Node · C++ · SQL AI & security agent guardrails · RAG · MCP · policy engines · reachability analysis · OpenVEX · SARIF · Ed25519 Backend FastAPI · Flask · Node · REST · WebSockets · JWT/OAuth2 · SQLAlchemy · Alembic Cloud & data Azure OpenAI · Azure AI Search · Google Cloud Run · BigQuery · Docker · PostgreSQL

🌐 moaz06.com · 💼 LinkedIn · ✉️ mohamedazahrioui2006@gmail.com

Pinned Loading

  1. FetchGate FetchGate Public

    Deterministic fetch-boundary gate for AI agents. a 200 is transport success, not a read. Python, MCP, stdlib-only.

    Python 3

  2. ReachGate ReachGate Public

    Deterministic vulnerability-reachability triage on GitLab Orbit's code graph: walks entry points to vulnerable code, exports OpenVEX/SARIF, verifies offline.

    Python

  3. TrustGate TrustGate Public

    Gemini agent action firewall using Fivetran and BigQuery evidence on Google Cloud Run.

    JavaScript

  4. bol-dpd-export bol-dpd-export Public

    Python

  5. procurement-risk-api procurement-risk-api Public

    Python

  6. algorithms-backtracking-champions-league algorithms-backtracking-champions-league Public

    Backtracking algorithm for generating valid competition schedules under multiple constraints (C++)

    C++