The AI-Powered Intelligent Incident Response Platform
Reimagining site reliability through real-time anomaly detection and LLM-driven diagnostics.
SentinelOps is a state-of-the-art AIOps platform designed to automate the entire SRE lifecycle. It acts as an intelligent layer above your infrastructure, ingesting raw logs and metrics to pinpoint anomalies before they become outages. By utilizing Large Language Models (LLMs), SentinelOps doesn't just tell you that something is wrong—it explains why and tells you how to fix it.
SentinelOps is a distributed microservices platform designed for automated incident detection and response. It leverages an event-driven Kafka pipeline to ingest logs and metrics in real time, apply AI-assisted root cause analysis, and serve intelligent alerts via a Next.js observability dashboard.
This project serves as a comprehensive portfolio piece demonstrating:
- Microservices Architecture
- Event-Driven Systems
- Distributed System Design
- Observability & Incident Management
- AI-assisted Root Cause Analysis
- Log ingestion pipeline
- Metrics monitoring
- AI anomaly detection
- Root cause analysis
- LLM-generated explanations
- Slack / Email alerting
- Observability dashboard
SentinelOps operates on a highly scalable, event-driven backbone utilizing Apache Kafka for asynchronous microservice orchestration.
graph TD
%% Source
L[Logs] --> LI(Log Ingestion)
M[Metrics] --> MC(Metrics Collector)
%% Ingestion
LI --> KLR(Kafka: logs.raw)
MC --> KMR(Kafka: metrics.raw)
%% Detection
KLR --> AD(Anomaly Detection)
KMR --> AD
AD --> KAD(Kafka: anomalies.detected)
%% Management
KAD --> IM(Incident Management)
IM --> KIC(Kafka: incidents.created)
%% Intelligence
KIC --> RCA(Root Cause Analysis)
RCA --> KIA(Kafka: incidents.analyzed)
KIA --> LLM(LLM Explanation)
LLM --> KIE(Kafka: incidents.explained)
%% Output
KIE --> AS(Alert Service)
AS --> SL(Slack / Email)
%% Dashboard
IM -.-> DASH(Next.js Frontend)
Each layer of SentinelOps is decoupled for maximum maintainability:
services/→ backend microservicesfrontend/→ Next.js dashboardinfrastructure/→ Docker / Kubernetes / Terraform configsconfigs/→ shared configurationai-models/→ AI componentsdocs/→ architecture documentation
- Backend: Python, FastAPI, Pydantic, Uvicorn
- Messaging: Apache Kafka, Zookeeper
- Frontend: Next.js 14, React, Framer Motion, Tailwind CSS
- Infrastructure: Docker, Docker Compose
cd infrastructure/docker
docker-compose up -dNavigate to services/<service-name>, create a virtual environment, install requirements, and run:
uvicorn src.main:app --port 800X --reloadcd frontend/incident-dashboard
npm install
npm run devSentinelOps includes a pre-configured monitoring stack using Prometheus and Grafana:
- Prometheus: Scrapes metrics from
log-ingestion-service,metrics-collector-service,anomaly-detection-service, andincident-management-service. - Grafana: Provides a visual dashboard for monitoring service health and metrics. Configuration is located in
infrastructure/monitoring/.
A robust Continuous Integration and Deployment pipeline is implemented using GitHub Actions (.github/workflows/ci.yml). The pipeline ensures:
- Automated dependency installation for all services.
- Code quality checks via
flake8linting. - Automated test execution.
- Docker image builds for the frontend and microservices.
SentinelOps is built for cloud-native orchestration. Kubernetes manifests are provided in infrastructure/kubernetes/ for key services:
log-ingestion-servicemetrics-collector-serviceanomaly-detection-serviceincident-management-service
Each service is defined with a deployment.yaml for self-healing scaling and a service.yaml for cluster-internal discovery.
- Prometheus monitoring
- Grafana dashboards
- Kubernetes deployment
- Distributed tracing
- CI/CD pipeline with GitHub Actions
Built with ❤️ by Ranjithhub08