Skip to content

AIOps Tooling and Stack

First PublishedByAtif Alam

This page is a practical reference for tools and libraries mentioned in AI-assisted SRE contexts. By the end, you should be able to pick a minimal stack for a pilot and justify tradeoffs.

AIOps and intelligent observability (platforms)

Section titled “AIOps and intelligent observability (platforms)”

Know conceptually (not every feature):

AreaExamples
APM / monitoring AIDatadog Watchdog, Dynatrace Davis
AWSDevOps Guru
Incident correlationMoogsoft, BigPanda, PagerDuty AIOps

Anomaly detection (Python)

Section titled “Anomaly detection (Python)”
LibraryTypical use
prophetTime-series forecasting (capacity, traffic)
scikit-learnIsolationForest, clustering for alert grouping
statsmodelsStatistical tests, ARIMA
pyodOutlier detection
riverOnline/streaming models

RAG and LLM integration

Section titled “RAG and LLM integration”
LayerExamples
OrchestrationLangChain
RAG indexingLlamaIndex
APIsopenai, anthropic SDKs

Embeddings and vector stores

Section titled “Embeddings and vector stores”
ComponentExamples
Embeddingssentence-transformers
Local / prototypeChromaDB, FAISS
ManagedPinecone, Weaviate

Intelligent runbooks and coding assistants

Section titled “Intelligent runbooks and coding assistants”
ToolNotes
CursorMulti-file context
AiderTerminal coding agent
Continue.devOpen-source, self-hostable

Log parsing and diagnostics

Section titled “Log parsing and diagnostics”
LibraryNotes
loguruStructured logging
elasticsearch-pyQuery logs for LLM context
tiktokenToken counting for context limits

Structured output and incident summaries

Section titled “Structured output and incident summaries”
LibraryNotes
instructorPydantic-structured LLM outputs
guidance (Microsoft)constrained generation

Evaluation

Section titled “Evaluation”
ToolNotes
ragasRAG faithfulness, relevance
deepevalTest-style assertions
promptfooPrompt regression across models

Experimentation and drift (optional)

Section titled “Experimentation and drift (optional)”
ToolNotes
mlflowExperiments, model registry
evidentlyDrift monitoring
great-expectationsData quality checks

Highest-ROI practice exercises

Section titled “Highest-ROI practice exercises”
  1. LangChain + ChromaDB + Anthropic — small RAG over runbooks.
  2. Prophet or IsolationForest — anomalies on sample metrics.
  3. Instructor + Anthropic — structured fields from a log snippet.
  4. Ragas — evaluate the RAG pipeline.
Section titled “Related reading”