Raphael Benoit

Senior advisor · Cybersecurity, AI & automation

I build intelligent, secure and automated systems.

Trained in software engineering in Montréal, I develop end-to-end products and tools across applied AI, Python automation, full-stack software and offensive security.

Experience

From industrial systems to enterprise cybersecurity.

Analysis, design, development, technical evidence and remediation guidance: I take ownership of the complete cycle.

Jul. 2026 — present

Senior cybersecurity & AI advisor

CGI · Montréal

Offensive security, Active Directory hardening, Python automation and applied AI in regulated environments.

Identity and access: Python/LLM tools used by multiple teams to restructure and produce technical documentation.
Active Directory security: risk analysis, permissions, groups, delegations and remediation across environments.
Web penetration testing: vulnerability validation, technical evidence, risk assessment and actionable fixes.
Oct. 2024 — Jun. 2026

Cybersecurity & AI consultant · DevSecOps

CGI · Montréal

IBM i/AS400 exfiltration-risk analysis, CyberArk/Power BI automation, SAST auto-remediation and penetration-test reporting tools.

Feb. 2019 — Apr. 2024

Programmer & technical lead

Letico · Saint-Laurent

Led Python, industrial automation, IT integration and internal security projects from requirements through maintenance.

Loading tracking: weight sensors, ERP/SAP, Web API, database and access control.
Production lines: PLCs, CIP, Modbus TCP/IP, sockets, statistics and anomaly detection.
Real-time systems: Python events and alerts connected to industrial controllers and databases.

Selected projects · 01

AI, products & automation.

Computer vision, full-stack products, local LLMs, automation and demonstrable industrial systems.

YOLO Object Detection Toolkit polygon annotation interface
Vision · Python · Open source

YOLO Object Detection Toolkit

Data preparation, assisted annotation, reproducible versions and visual review.

Case study
Vision · Segmentation

Plant detection and segmentation

Boxes, polygons and segmentation under occlusion in an anonymized demonstration.

TheJobApplier dashboard
SaaS · Chrome MV3 · AI

TheJobApplier

Full-stack product: async Quart, OAuth, Stripe, AWS and a published extension.

Case study
Multi-GPU local AI and cybersecurity server assembled by Raphael Benoit
AI infrastructure · Multi-GPU · Local LLM

Local AI lab · 120 GB VRAM

Server designed and assembled for LLMs, RAG, fine-tuning, vision and cybersecurity workloads.

View project
Local LLM · RAG

Private document assistant

File ingestion, contextual search and fully local operation.

STT · LLM · Word

Automated professional reports

Conversation and form inputs into a structured, reviewable report.

Cybersecurity · RAG

Technical document research

Search and synthesis in a specialized document collection.

Web · AI assistant

Documentation-grounded Web assistant

Answers grounded in selected sources within a Web interface.

Audio · Transcription

Transcription and speakers

Timestamps and speaker separation in a complete workflow.

Industrial automation connected to production equipment
PLC · ERP · Python

Industrial automation

Sensors, SAP/ERP, CIP/Modbus, real-time events and anomaly detection.

Selected projects · 02

Cybersecurity, pentesting & research.

Offensive tools, reverse engineering, Windows kernel work and DevSecOps automation, with public studies and demonstrations.

Agentic auto-pentest architecture
Pentest · AI agents · Proprietary

Agentic auto-pentest

Orchestration with memory, tools, scope enforcement, evidence and human validation.

Case study
Defensive Ghidra and LLM pipeline
Reverse engineering · Python · Open source

Ghidra LLM Reverse Pipeline

Compact binary exports, configurable LLM analysis and preserved provenance.

Case study
Windows internals research from user mode to kernel
4 articles · C++ · HTB Hard challenge

From user mode to ring 0

Memory analysis, hooking, Windows drivers, WinDbg, page tables and detection surfaces.

Case study
AI-assisted DevSecOps pipeline
DevSecOps · SAST · LLM

AI applied to security

SAST auto-remediation, assisted fixes and automated penetration-test reporting.

View public tools

Selected demonstrations

Antivirus and kernel research first, followed by the strongest internal implementations.

A selection of the most representative demonstrations. Explore the 11-video catalog →

Contact & profiles

Raphael Benoit — cybersecurity, AI and software engineering.

LinkedIn summarizes the professional path; GitHub, HackTheBox and YouTube provide the projects, rankings and technical demonstrations.