🔭 03: The AI Observatory#
Welcome to the third adventure in the Open Ecosystem Challenge series! Your mission: investigate a mysterious bandwidth anomaly at a remote research station by instrumenting its AI system. This is a hands-on journey through AI Observability with OpenTelemetry, OpenLLMetry, and Jaeger.
The entire infrastructure is pre-provisioned in your Codespace — Kubernetes cluster, Ollama, and observability tools are ready to go. You don't need to set up anything locally. Just focus on solving the problem.
🪐 The Backstory#
You are stationed at Perimeter Alpha, a research outpost on the newly discovered planet HB-7742. The station is run by HubSystem, a central AI that manages everything from life support to data analysis.
Recently, the station's bandwidth usage has spiked to 847% above baseline, but no one knows why. As the systems engineer, it's your job to instrument the AI, trace its activities, and uncover the root cause of the anomaly.
Your mission: Bring visibility to the station's AI and solve the mystery.
🎮 Choose Your Level#
Each level is a standalone challenge with its own Codespace that builds on the story while being technically independent — pick your level and start wherever you feel comfortable!
💡 Not sure which level to choose? Learn more about levels
🟢 Beginner: Calibrating the Lens#
- Status: ✅ Available
- **Topics: ** OpenTelemetry, OpenLLMetry, Jaeger
The HubSystem is running "blind". Your mission: instrument the Python application with OpenLLMetry, send traces to the collector, and use Jaeger to find out what the AI is actually doing.
🟡 Intermediate: The Distracted Pilot#
- Status: ✅ Available
- **Topics: ** OpenTelemetry, OpenLLMetry, Jaeger, Prometheus
You've escaped aboard the Perihelion, a research vessel piloted by a very opinionated AI called ART. The jump coordinates to RaviHyral should have been ready an hour ago — but ART is distracted. Your mission: instrument the RAG pipeline, track what ART is actually retrieving, and fix the navigation system before you miss the jump window.
Start the Intermediate Challenge
🔴 Expert: The Noise Filter#
- Status: ✅ Available
- **Topics: ** OpenTelemetry, OpenTelemetry Collector, Jaeger
You made it to RaviHyral. ART offered to share its observability data with the local station — but the traces are a mess. Non-standard span names, missing token usage, and Jaeger drowning in noise. Your mission: fix ART's instrumentation to follow GenAI semantic conventions, record errors properly, and configure tail sampling to filter out the noise.