Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: MLOps
Why Agent Accuracy Scores Mislead: pass^k, Perturbations and Fault Injection
Why pass@1 hides unreliable AI agents, and how pass^k, input perturbation sets and fault injection reveal consistency, stability and fault-tolerance gaps.
9 min readTrain a Custom Model in Azure ML and Call It from a Foundry Agent
Train, register and deploy a churn model on an Azure ML managed endpoint, then expose it to a Foundry agent as a FunctionTool, with auth and production notes.
7 min readBefore an AI Agent Gets Production Authority: Contracts, Claim Graphs, Shadow Runs and Canaries
An evidence-first release process for AI agents: scoped contracts, stateful scenarios, fault injection, uncertainty bounds, staged canaries and machine-enforced promotion.
23 min readFrom Models to Agents: Reworking the Three Lines of Defense
How MLOps governance maps onto agent systems: what each of the three lines must add, where the gaps sit, and what EU, FDA and finance regulators signal.
8 min readDeploying AI Agents to Production: Cloud, Self-Hosted or Hybrid Trade-offs
Compare cloud, self-hosted and hybrid agent deployments on data exposure, cost, lock-in, routing and observability, with four questions to choose and mistakes to avoid.
10 min readA LangGraph Pipeline That Keeps Codebase Memory for AI Agents Accurate
Build a LangGraph pipeline that drafts per-service knowledge files, gates them with Ragas and DeepEval, traces runs in Langfuse and runs locally on Ollama too.
11 min readShadow, Canary, A/B and Bandits: Rolling Out ML Models Safely
A fraud-detection example shows when to use shadow deployment, canary releases, A/B tests and multi-armed bandits, what to monitor and how to roll back.
7 min readModel Serving Architecture: Why the API Is Only the Front Door
A guide to the platform behind an inference endpoint: routing, scaling, versioning, observability, security, and calling served models from Spring AI or LangChain4j.
9 min readPractical notes: Designing a Cache for AI Agents: Running It in Production
Operable walkthrough of Practical notes: Designing a Cache for AI Agents: Running It in Production: contracts, checks, and drop-in code slots for teams shipping this pattern.
7 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.