SentinelAI – DevOps Security Agent

“SentinelAI detecting intrusion in real-time and automatically triggering Telegram alerts and GitLab issue creation.”

SentinelAI – DevOps Security Agent 💡 Inspiration

With increasing security concerns in public and private spaces, traditional CCTV systems only record incidents but do not actively prevent threats.

We wanted to build a system that acts as an intelligent security assistant — capable of detecting suspicious activity in real-time and responding instantly.

🚀 What it does

SentinelAI is an AI-powered surveillance and DevOps security agent that:

Detects intrusions, unauthorized access, and suspicious behavior Sends instant alerts via Telegram Automatically creates GitLab issues for incident tracking Automates security incident response workflows Provides a real-time monitoring dashboard Logs events for analysis and future prevention

👉 It transforms CCTV into an active AI-driven protection system

🔗 GitLab DevOps Integration (AI Agent)

SentinelAI extends beyond surveillance by acting as an AI-powered DevOps security agent.

When a threat is detected, the system can:

🛠️ Automatically create a GitLab issue 🚨 Assign severity labels (LOW, MEDIUM, HIGH, CRITICAL) 📎 Attach captured image/video as evidence 🔄 Enable teams to track and respond to incidents within their DevOps workflow

👉 This bridges physical security monitoring with Software Development Lifecycle (SDLC) operations

🔥 SentinelAI converts real-world security events into automated DevOps workflows using an event-driven AI agent architecture.

🧠 How we built it

We combined Computer Vision + AI + IoT + Web technologies:

Object Detection Model: YOLO (Ultralytics) Backend: Python + Flask Computer Vision: OpenCV Alerts: Telegram Bot API DevOps Integration: GitLab API Frontend: HTML, CSS, JavaScript Camera Input: Webcam/IP camera 🔄 System Workflow 📷 Camera captures live video feed 🤖 AI model detects objects and suspicious behavior ⚠️ Threat is identified 📲 Telegram alert is sent 🛠️ GitLab issue is automatically created 📊 Data is logged in the dashboard ⚙️ Challenges we ran into ⚡ Real-time processing without lag 🎯 Reducing false positives 🔌 Integrating hardware with AI pipeline 📡 Ensuring reliable alert delivery 🔗 Mapping physical events to DevOps workflows 🏆 What we learned Real-world Computer Vision deployment Building end-to-end AI systems API integration (Telegram + GitLab) Importance of security + usability in AI systems 🌍 Impact

SentinelAI can be used in:

🏠 Homes 🏫 Schools 🏢 Offices 🏭 Industrial infrastructure

👉 It connects physical security with DevOps workflows, enabling proactive incident response.

🤖 GitLab Duo Agent Architecture

SentinelAI is built as an event-driven AI agent compatible with the GitLab Duo Agent Platform.

Trigger: Real-time security event (intrusion / loitering detection) Context: Captured image, timestamp, and threat level classification Action: Automatically creates and updates GitLab issues with severity labels and evidence

This enables automated incident response workflows inside GitLab, reducing manual monitoring and improving SDLC security operations.

⚙️ AI Agent Workflow Event Triggered → Camera detects suspicious activity AI Processing → YOLOv8 identifies intrusion/loitering Context Generation → Threat level + media evidence prepared Agent Action → GitLab issue automatically created Notification → Telegram alert sent Output → Incident tracked in DevOps workflow 🔐 Relevance to SDLC

SentinelAI reduces manual security monitoring by automating incident detection and tracking directly inside GitLab workflows, improving operational efficiency and response time.

🎯 Conclusion

SentinelAI demonstrates how AI can go beyond monitoring and become an intelligent DevOps security agent, enabling real-time threat detection and automated incident management.