Distributed Logging Service
Staff-level design for a distributed logging platform supporting wildcard querying, custom aggregations at 1-minute granularity, and 1M+ events/sec ingest.
6 docs tagged with "observability"
View all tagsFeature drift — what it is, how it differs from data drift, and how to monitor it
A practitioner's guide to feature drift in production ML systems, covering definitions, the relationship to data drift, monitoring metrics, and response thresholds.
Feature Drift Detection: Staff-Level Ownership and Execution Story
A detailed interview narrative for explaining how I led feature drift detection from ambiguous product needs to a scalable production system.
ML Observability — Core Concepts
Data drift, concept drift, metrics, telemetry, and the three pillars of ML observability.
ML Observability — Debugging Production Models & Why It's Different
A practitioner's guide to diagnosing underperforming ML models in production and understanding why ML observability is a distinct discipline from traditional software monitoring.
Model Foundry AI Agent for Observability Q&A and Metric Diagnosis
Staff-level system design for a conversational AI agent that reuses Model Foundry metrics APIs to answer questions, detect unhealthy models, explain anomalies, and recommend approved actions.