This collection is a selection of AI-driven solutions I developed for cybersecurity and security intelligence applications, among other use cases. It features AI agents, API integrations, and customizable models designed for seamless compatibility with modern threat management systems.
These projects demonstrate the integration of artificial intelligence with cybersecurity, offering capabilities such as real-time threat detection and advanced analytical tools to strengthen security infrastructure.
This portfolio demonstrates my proven ability and expertise to deliver AI solutions—from design and development to practical real-world implementation.
Tip
Live Demos Available
Several projects are hosted online—click the badge below to explore them:
Note for project "E. 🕵️ Security Image & Video Analyzer"
This Python-based solution requires a compatible AI model. Download it here:
Download Model
After downloading, rename the file to yolo11x.pt for seamless integration.
- 🌟 Featured Projects
A. 🔒 Advanced Threat Video Analyzer
B. 📋 AI Compliance Visual Inspector
C. 🤖 Security AI Agent
D. 🛠️ Multi-Tool AI Assistant
E. 🕵️ Security Image & Video Analyzer
F. 🐉 Chinese Website Translator - Chrome AI API Extension
G. 📚 Person and Object Detection - 🚀 Getting Started
- 📂 Repository Structure
- 🤝 Contribution & Feedback
Updated: 31/08/2025
🚧 Expected Release: Q4 2025 - Q1 2026
Next-generation security vision system featuring:
- Real-time Object detection
- Animals and person detection
- Behavioral pattern analysis
Note: Try beta version online here
A Vision AI Agent live for security cameras to:
- Policy violation detection:
- Unattended devices (phones, laptops)
- Clean-desk violations
- Employees/Visitors without badges
- Paper, pen, pencil, unathourized devices at production
- Open or damaged doors/windows
- Any other clean desk policy your company have
This application adds that context by:
-
Framing the tool for security and compliance use cases (e.g., detecting unattended devices, clean-desk issues).
-
Providing a confidence threshold to help filter out uncertain results, which is critical for making informed decisions.
The interpretation of the AI's output is left to the human user. For example, if the AI identifies a laptop, monitor, and keyboard on a desk, the user must decide if this scene violates a specific clean-desk policy. This tool acts as a powerful assistant to flag potential issues, not as an autonomous decision-maker.
Important
- This tool is a prototype and demonstration. It is not a certified compliance auditing tool.
- The AI can and will make mistakes—it might miss objects (false negatives) or misidentify (like the example image above that it detected a notebook as a "desktop computer") them (false positives).
- Always have a human expert review and validate any findings before making any compliance or security-related decisions.
- Do not rely solely on this tool's output.
An AI Agent that decides wich tool to use to analyze and select cybersecurity tools/apps and report.
This project is under development and cybersecurity tools are beeing made with Python compatible with the agent.
The project is expected to be completed by the end of Q4
A Template from Hugging Face Agent Course
Core Capabilities:
- Time zone conversion engine
- Real-time weather lookups
- Unit conversion system
- Intelligent web search
- Dynamic image generation
- Python code execution
Example Use Cases:
"What time is 3pm EST in Tokyo?"
"Show weather in New York"
"Convert 50 miles to kilometers"
"Generate image of secure server room"Technical Highlights:
- Qwen2.5-Coder-32B model backbone
- Multi-step reasoning architecture
- Tool chaining capabilities
- Hugging Face optimized deployment
A Desktop Python-based tool for automated image and video analysis using advanced object detection.
Designed for security professionals to enhance incident investigation, forensic analysis, and threat detection workflows.
Important
You need a pre-trained model in the same folder of the project to run this code.
A powerful Chrome AI API browser extension that translates Chinese websites (both simplified and traditional) into English (US) with a stunning dark, futuristic interface design.
Download the browser extension click here
Features:
🌟 Core Functionality
- Dual Translation APIs: Uses Chrome's built-in AI translator with Google Translate fallback
- Full Page Translation: Instantly translate entire Chinese web pages to English
- Hover Translation: Real-time translation tooltips when hovering over Chinese text
- Language Detection: Automatically detects simplified vs traditional Chinese
What it does:
- Smart Detection: Identifies people, animals, and everyday objects in images.
- Visual Highlights: 🟨 Draws precise bounding boxes 🟩 around detected targets 🟥—so you see exactly what the AI found.
- Powered by YOLOv11: Trained on COCO128 for reliable, real-world performance.
Perfect for: Security checks, wildlife monitoring, retail inventory, or just exploring AI vision!
🚀 Try the demo now—upload any image and watch the AI in action!
-
For live demos:
- Click the demo badges above
- Allow 20-30 seconds for Hugging Face to load
- Interact with the interfaces
-
For under development projects:
- Check back for Q4 2025 release
- Contact for collaboration opportunities
AI-Security-Projects/
├── README.md # Project documentation
├── security-image-analyzer # Has 2 images, its a simple AI image detector made with python
├── images # Images of the projects
└── chrome_extension_chinese_translator_web # Chrome AI API browser extension
This repository demonstrates practical AI security implementations. Suggestions and use case ideas are welcome!
Let's chat:
💼 LinkedIn: https://www.linkedin.com/in/thiago-cequeira-99202239/
🧑💻 GitHub: https://github.com/ThiagoMaria-SecurityIT/
🤗 Hugging Face: https://huggingface.co/ThiSecur