| { |
| "name": "Production AI Observability Monitor", |
| "problem": "Production AI teams need trace-level signals for latency, token growth, tool failures, and low-quality outputs.", |
| "domain": "ai-observability", |
| "architecture": "classifier", |
| "hugging_face_tasks": [ |
| "text-classification", |
| "token-classification", |
| "summarization", |
| "zero-shot-classification" |
| ], |
| "recommended_stack": [ |
| "OpenTelemetry GenAI semantic conventions", |
| "OpenTelemetry Collector for vendor-neutral ingestion", |
| "ClickHouse or PostgreSQL for trace analytics", |
| "Prometheus and Grafana for service-level metrics", |
| "MLflow for model and prompt version linkage", |
| "FastAPI for trace search and regression APIs" |
| ], |
| "real_world_data_sources": [ |
| { |
| "name": "Prometheus HTTP API", |
| "url": "https://prometheus.demo.do.prometheus.io/api/v1/query?query=up", |
| "purpose": "Real time-series telemetry for anomaly pipelines" |
| }, |
| { |
| "name": "GitHub Events API", |
| "url": "https://api.github.com/events?per_page=10", |
| "purpose": "Deployment-correlated public event metadata" |
| } |
| ], |
| "job_description_skills": [ |
| "LLMOps observability and trace instrumentation", |
| "Prompt, model, dataset, and deployment lineage", |
| "SLOs, anomaly detection, and incident diagnostics", |
| "High-volume telemetry storage and aggregation", |
| "Evaluation-driven production monitoring" |
| ], |
| "impact_targets": [ |
| "Ingest 1,000 synthetic traces/second without loss", |
| "Detect seeded latency and quality regressions with >= 0.90 precision", |
| "Link 100% of traces to model, prompt, and dataset versions", |
| "Generate a release health report in under 60 seconds" |
| ], |
| "baseline_evaluation": { |
| "test_examples": 4, |
| "accuracy": 1, |
| "synthetic_evaluation": true |
| }, |
| "estimated_delivery": "8-12 weeks for one engineer", |
| "generated_baseline_is_production_ready": false |
| } |
|
|
