Developer Articles | TechForDev

Latest AI / ML JavaScript Python React Next.js Web Dev DevOps Cloud

CDEvents in Action #7: Instrument Any CI Step in a Few Lines

David BernardApr 13, 2026 • 3 min read

CDEvents in Action #7: Instrument Any CI Step in a Few Lines

Webhook integrations (ep#3, ep#4) tell you when a pipeline started and whether it passed. They don't...

#testing#observability#devops#cicd

0 0

Closing the Eval Gap: From Lenient Defaults to Signal That Matters

Ian Parent5d ago • 5 min read

Closing the Eval Gap: From Lenient Defaults to Signal That Matters

In the original Eval Gap post, we laid out the problem: the distance between "works in demo" and...

#mcp#aiagents#observability#opensource

1 0

The $47,000 Agent Loop: What Observability Would Have Caught on Day One

Gabriel Anhaia2d ago • 5 min read

The $47,000 Agent Loop: What Observability Would Have Caught on Day One

Four LangChain agents ran for 11 days on a misclassified retry. $47K bill. Every span returned 200. ...

#ai#observability#llm#devops

0 1

Langfuse vs LangSmith vs Phoenix vs Braintrust: The Honest 2026 Comparison

Gabriel Anhaia2d ago • 5 min read

Langfuse vs LangSmith vs Phoenix vs Braintrust: The Honest 2026 Comparison

Four LLM observability platforms, one real workload, no vendor angle. What each gets right, what it ...

#ai#observability#llm#devops

0 0

Structure-Driven Engineering Organization Theory #1 — The Concept of Observation

machuz3d ago • 9 min read

Structure-Driven Engineering Organization Theory #1 — The Concept of Observation

Evaluation assigns values. Observation reads structure. Get the entry point wrong and every destinat...

#management#leadership#engineering#observability

0 0

[The 8-Turn Problem] Why Your Agent Fails at Turn 3 and You Only Notice at Turn 7

Tijo Gaucher14h ago • 5 min read

[The 8-Turn Problem] Why Your Agent Fails at Turn 3 and You Only Notice at Turn 7

Last Tuesday an agent I shipped decided, mid-conversation, that the user's name was "Export CSV." It...

#ai#observability#devops#webdev

0 0

The Failure Mode Your Observability Stack Cannot See

Gabriel Anhaia2d ago • 6 min read

The Failure Mode Your Observability Stack Cannot See

02:14. Every span in the trace is green. The customer says the AI is lying. Both are true. What your...

#ai#observability#llm#devops

0 0

Datadog Sees the HTTP 200. It Cannot See the Hallucination.

Gabriel Anhaia2d ago • 4 min read

Datadog Sees the HTTP 200. It Cannot See the Hallucination.

Your APM is world-class at HTTP. It has no vocabulary for a response that is fluent, confident, and ...

#ai#observability#llm#devops

0 0

Your LLM Gateway Is a Blind Spot. Here's How to Instrument It After the LiteLLM Incident.

Gabriel Anhaia2d ago • 5 min read

Your LLM Gateway Is a Blind Spot. Here's How to Instrument It After the LiteLLM Incident.

The gateway in your stack routes every LLM call. After March 2026's LiteLLM security incident, treat...

#ai#observability#security#devops

0 0

CloudWatch literacy: the QA superpower that routes bugs without triage theater

Aman Bhandari1d ago • 6 min read

CloudWatch literacy: the QA superpower that routes bugs without triage theater

Triage theater is the 40-minute meeting that starts with QA saying "users report the upload is...

#qa#aws#devops#observability

0 0

Debugging an LLM Bug at 3 AM: The Runbook I Wish I'd Had

Gabriel Anhaia2d ago • 9 min read

Debugging an LLM Bug at 3 AM: The Runbook I Wish I'd Had

A time-based playbook for on-call engineers responding to an LLM-feature incident. Five triage branc...

#ai#observability#devops#llm

0 0

Your LLM Got Quietly Dumber Last Week. Your Dashboards Have No Idea.

Gabriel Anhaia2d ago • 5 min read

Your LLM Got Quietly Dumber Last Week. Your Dashboards Have No Idea.

Opus 4.6 users are reporting a sharp quality drop. Anthropic isn't saying. Your APM sees nothing. He...

#ai#observability#llm#devops

0 0

A Review of 100 Public AI Incident Postmortems. Here Are the 6 Mistakes That Keep Showing Up.

Gabriel Anhaia2d ago • 9 min read

A Review of 100 Public AI Incident Postmortems. Here Are the 6 Mistakes That Keep Showing Up.

Read every public AI-incident postmortem from the last two years and six patterns repeat. Six. One i...

#ai#observability#llm#devops

0 0

8 Ways Your LLM App Is Silently Failing Right Now (and What to Instrument for Each)

Gabriel Anhaia2d ago • 6 min read

8 Ways Your LLM App Is Silently Failing Right Now (and What to Instrument for Each)

Every failure mode here returns 200. Latency is fine. Cost looks reasonable. Users are furious. Here...

#ai#observability#llm#devops

0 0

The 7 Most Expensive LLM Production Incidents of 2025–2026 (Each One Had a Fixable Signal Nobody Watched)

Gabriel Anhaia2d ago • 11 min read

The 7 Most Expensive LLM Production Incidents of 2025–2026 (Each One Had a Fixable Signal Nobody Watched)

Seven real LLM outages with dollar figures attached. For each: the signal that would have caught it ...

#ai#observability#llm#devops

0 0

OpenTelemetry GenAI Semantic Conventions: Your LLM Traces Should Look Like This in 2026

Gabriel Anhaia2d ago • 5 min read

OpenTelemetry GenAI Semantic Conventions: Your LLM Traces Should Look Like This in 2026

The vendor-neutral standard for LLM tracing. What attributes to emit, what the span tree should look...

#ai#observability#opentelemetry#llm

0 0

The Production Readiness Checklist for LLM Apps Nobody Gave You (18 Items)

Gabriel Anhaia2d ago • 5 min read

The Production Readiness Checklist for LLM Apps Nobody Gave You (18 Items)

Before your LLM feature meets a paying customer, these 18 items should all be true. Copy-paste, tick...

#ai#observability#llm#devops

0 0

The 5 RAG Failure Modes Nobody Talks About (and How to Detect Them Before Users Do)

Gabriel Anhaia2d ago • 8 min read

The 5 RAG Failure Modes Nobody Talks About (and How to Detect Them Before Users Do)

Your retrieval metrics are green. Your users are confused. Five RAG failure modes that aggregate das...

#ai#llm#rag#observability

0 0

Claude Code Charges You and Won't Tell You Why. The Community Fixed It

Nirav Joshi6d ago • 5 min read

Claude Code Charges You and Won't Tell You Why. The Community Fixed It

Claude Code logs everything but surfaces nothing. Three developers built the observability layer the...

#ai#aiagents#observability#programming

0 0

LLM-as-Judge: The Eval Technique That Looks Cheap Until It Grades Its Own Bias Back to You

Gabriel Anhaia2d ago • 7 min read

LLM-as-Judge: The Eval Technique That Looks Cheap Until It Grades Its Own Bias Back to You

A team ran LLM-as-judge for 11 weeks with a green dashboard. The judge had learned to admire its own...

#ai#testing#llm#observability

0 0

Your MCP server just started telling on itself (in a good way)

Dinesh Kumar6d ago • 3 min read

Your MCP server just started telling on itself (in a good way)

The gap nobody talks about There are 10,000+ Model Context Protocol servers now. Every...

#mcp#python#observability#aiagents

0 0

What Happens When an AI Agent Gets Kernel-Level GPU Traces

Ingero Team4d ago • 5 min read

What Happens When an AI Agent Gets Kernel-Level GPU Traces

TL;DR A GPU trace of a PyTorch DataLoader bottleneck (114x slower than direct indexing)...

#gpu#ebpf#observability#gpuobservability

0 0

From $50K/Year of Datadog to $0/Year of Self-Hosted Observability: The Migration Every Team Is Doing in 2026

Gabriel Anhaia2d ago • 10 min read

From $50K/Year of Datadog to $0/Year of Self-Hosted Observability: The Migration Every Team Is Doing in 2026

Your Datadog bill crossed $50K and finance flagged it. Here is the 2026 migration path teams are act...

#devops#observability#opentelemetry#llm

1 0

Claude Went Down Twice in 48 Hours Last Week. If You Noticed, Your Fallback Failed.

Gabriel Anhaia2d ago • 8 min read

Claude Went Down Twice in 48 Hours Last Week. If You Noticed, Your Fallback Failed.

Anthropic had two outages on April 7 and 8, 2026. If your users felt them, your multi-provider fallb...

#ai#devops#llm#observability

0 0

Tech Articles

CDEvents in Action #7: Instrument Any CI Step in a Few Lines

Closing the Eval Gap: From Lenient Defaults to Signal That Matters

The $47,000 Agent Loop: What Observability Would Have Caught on Day One

Langfuse vs LangSmith vs Phoenix vs Braintrust: The Honest 2026 Comparison

Structure-Driven Engineering Organization Theory #1 — The Concept of Observation

[The 8-Turn Problem] Why Your Agent Fails at Turn 3 and You Only Notice at Turn 7

The Failure Mode Your Observability Stack Cannot See

Datadog Sees the HTTP 200. It Cannot See the Hallucination.

Your LLM Gateway Is a Blind Spot. Here's How to Instrument It After the LiteLLM Incident.

CloudWatch literacy: the QA superpower that routes bugs without triage theater

Debugging an LLM Bug at 3 AM: The Runbook I Wish I'd Had

Your LLM Got Quietly Dumber Last Week. Your Dashboards Have No Idea.

A Review of 100 Public AI Incident Postmortems. Here Are the 6 Mistakes That Keep Showing Up.

8 Ways Your LLM App Is Silently Failing Right Now (and What to Instrument for Each)

The 7 Most Expensive LLM Production Incidents of 2025–2026 (Each One Had a Fixable Signal Nobody Watched)

OpenTelemetry GenAI Semantic Conventions: Your LLM Traces Should Look Like This in 2026

The Production Readiness Checklist for LLM Apps Nobody Gave You (18 Items)

The 5 RAG Failure Modes Nobody Talks About (and How to Detect Them Before Users Do)

Claude Code Charges You and Won't Tell You Why. The Community Fixed It

LLM-as-Judge: The Eval Technique That Looks Cheap Until It Grades Its Own Bias Back to You

Your MCP server just started telling on itself (in a good way)

What Happens When an AI Agent Gets Kernel-Level GPU Traces

From $50K/Year of Datadog to $0/Year of Self-Hosted Observability: The Migration Every Team Is Doing in 2026

Claude Went Down Twice in 48 Hours Last Week. If You Noticed, Your Fallback Failed.