The problem

Teams need to show how monitoring, metrics, and early AIOps fit together—without running production systems. A coherent demo (microservices → Prometheus → anomaly detection → dashboards) is hard to spin up and explain in interviews or pitches.

The solution

SignalGuard is a full observability demo: synthetic FastAPI microservices emit Prometheus metrics, a separate anomaly-detection service scores them, Prometheus and Grafana are provisioned via Docker, and a standalone GitHub Pages dashboard simulates live metrics and anomalies—no backend required for the static demo.

Without SignalGuard

Ad-hoc scripts or production-only setups; no single story for “metrics + anomaly + dashboards.”

With SignalGuard

One repo: microservices, Prometheus, Grafana, anomaly detector, and a static demo dashboard—interview- and demo-ready.

What it does

  • Synthetic microservices – FastAPI apps with latency, error rate, throughput, and business metrics exported to Prometheus.
  • Anomaly detection service – Queries Prometheus, computes anomaly scores, emits flags and scores for dashboards.
  • Prometheus & Grafana – Scrape configs and prebuilt dashboards (latency p95, error rate, throughput, anomaly score).
  • GitHub Pages dashboard – Static, simulated metrics and anomalies with Chart.js; no backend; ideal for demos.

Tech stack

FastAPI, Prometheus, Grafana, Python (anomaly detector), Chart.js (static dashboard), Docker Compose.

Next steps

Roadmap & ideas

  • More anomaly methods (e.g. forecasting, baselines) and alerting rules.
  • Additional synthetic services and failure scenarios.
  • Optional integration with signalguard-aiops (log intelligence) for cross-signal correlation.
GitHub – signal-guard Live demo Back to projects