AI Engineer | PhD Researcher | Building the Future of Intelligent Security
My PhD research explores how Large Language Models can transform Network Intrusion Detection. The goal: AI security systems that genuinely understand threats rather than just matching patterns.
Current Focus:
| Area | Description |
|---|---|
| Research | LLM architectures for real-time threat detection and contextual analysis |
| Development | Nylah - an advanced AI assistant with genuine reasoning capabilities |
| Integration | Bridging cybersecurity fundamentals with production-ready AI systems |
| Methodology | Studying offensive security to build better defenses |
The premise is straightforward: effective defense requires understanding systems at a depth that traditional approaches cannot achieve. LLMs don't merely detect patterns; they can reason about context, intent, and anomaly. That capability is the next frontier of security, and I'm building toward it.
|
Languages & Core
|
Frameworks
|
|
LLMs & Models
|
RAG & Vector Systems
|
Computer Vision
Real-time Detection Facial Landmarks Pattern Recognition
|
Languages
|
Frameworks
|
Cloud & DevOps
|
RESTful APIs Microservices AWS SageMaker Lambda Serverless Architecture
| Domain | Capabilities |
|---|---|
| Network Security | Packet Analysis, IDS/IPS, Protocol Analysis |
| Application Security | Penetration Testing, Vulnerability Assessment, OWASP Top 10 |
| DevOps | CI/CD Pipelines, Container Orchestration, Infrastructure as Code |
| Automation | n8n Workflows, Process Orchestration, Integration Pipelines |
An advanced AI assistant designed to go beyond conversational interfaces. The architecture combines natural language understanding, autonomous task execution, and multi-modal processing with a coherent personality layer.
Core Capabilities:
- Contextual reasoning and memory persistence
- Task decomposition and autonomous execution
- Multi-modal input processing (text, voice, vision)
- Adaptive personality and communication style
Stack: Python LLMs RAG Vector DBs FastAPI Multi-modal AI
Status: Private repository. Public release planned.
|
Hybrid retrieval architecture combining dense and sparse methods for domain-specific knowledge injection. Focus on eliminating hallucinations while maintaining contextual depth. Objective: Retrieval that comprehends, not just matches.
|
Real-time computer vision system for driver safety monitoring. Implements facial landmark detection and eye-state analysis using optimized YOLO models. Application: Preventing fatigue-related accidents through proactive alerts.
|
|
AI-driven automation for job applications. Handles form parsing, intelligent field mapping, and submission across multiple platforms. Rationale: Repetitive tasks should be automated. Always.
|
End-to-end pipeline for domain adaptation of language models. Includes data quality validation, training optimization, and evaluation metrics. Focus: Ensuring models learn precisely what you intend.
|
|
Intelligent automation systems connecting productivity tools: Gmail, Telegram, Drive, Slack, and custom APIs. Orchestrated workflows for process optimization.
|
French to Fongbe translation system. Contributing to language preservation through accessible technology. Mission: Keeping languages alive in the digital era.
|
|
|
|
"The best security systems don't just react - they understand. The best AI doesn't just process - it reasons. I'm building both."
|
Learning in Public Sharing progress, including failures. |
Security Through Understanding Not obscurity. |
AI as Augmentation Extending human capability, not replacing it. |
Strategic Openness Open source when beneficial, closed when necessary. |
Sustained Focus Coffee is infrastructure. |
Building something interesting? Working on AI or security challenges? Open to collaboration and technical discussions.
| Contact | Address |
|---|---|
| Academic | m.wourougou@ueuromed.org |
| Personal | ravellewourougou@gmail.com |
| Phone | +212 645 469 843 |
| in/maako-wourougou | |
| Response Time | Typically within 24 hours |
Last updated: December 2025
In a world of pattern matchers, be the one who teaches machines to reason.