REST Observability in Azure API Management
Enable gateway logs, KQL, metrics, correlation IDs and alerts to spot latency, errors, and AI budget issues in Azure APIM.
Tag
MLOps
Articles tagged “MLOps” on Ryshe.
GxP Audit Trails for AI Workflows on Azure
Build Part 11/Annex 11‑compliant AI audit trails on Azure: log identity, model version, input/output hashes, immutable storage, retention and review.
Document-Level Translation Metrics: Comparison
Compare BLEU, COMET, BLEURT, BERTScore and discourse-aware metrics to evaluate long technical translations and spot cross-sentence errors.
Real-Time vs. Batch Feature Updates
Use real-time only when freshness changes decisions; otherwise prefer batch or a hybrid to cut cost, complexity, and audit risk.
Azure AI Quota Tracking for Shared Hubs
Track TPM, PTU, VM family and deployment quotas in shared Azure AI hubs using Foundry, Monitor, and Log Analytics; set alerts and owners.
Monitor Azure AI Inference with OpenTelemetry
Use OpenTelemetry and Application Insights to monitor Azure AI inference—track traces, p95 latency, token usage, errors, and audit tags from the first request.
Activity Duration Forecasting on Azure: Guide
Guide to building production-ready activity-duration forecasts on Azure: data pipelines, time-aware training, batch scoring, Power BI, and governance.
From AI Proof of Concept to Production: Why Most Projects Never Make It
Your AI proof of concept worked perfectly. So why is it still sitting in a notebook six months later? The gap between demo and production is where most AI investments go to die.