FLAGSHIP ROADMAP · FREE

AI for SRE: From AI Fundamentals to Production Reliability

AI for SRE is the practice of applying site reliability engineering thinking to AI systems where models, retrieval, agents, tools, GPUs and external providers all affect the user experience. This roadmap shows what an SRE, DevOps or platform engineer should learn — and in what order.

397 free conceptsNo uploadsNo login requiredProgress stays in your browser
THE SHORT ANSWER

What is AI for SRE?

Traditional SRE asks whether a service is available, fast enough, scalable and recoverable. AI systems add new dimensions: a request can return HTTP 200 while the answer is wrong, retrieval is stale, a tool fails, an agent loops, a provider throttles requests or GPU queues make the product feel unavailable. AI for SRE keeps the familiar reliability discipline, then extends it across the complete AI request path.

You do not need to become an ML researcher first. You need enough model knowledge to understand the system you operate, then you connect that knowledge to architecture, observability, capacity, SLOs, incident response, security and cost.

LEARNING ROADMAP

Learn the stack in the order an SRE operates it.

The fastest useful path is to build one mental model at a time. Start with what the model is, then learn the application layers around it, and only then go deep on production operations.

01

AI, ML and LLM fundamentals

Understand the difference between AI, machine learning, neural networks, transformers and large language models. Learn tokens, context, inference and why a model is only one component of the production system.

Start with AI →
02

Prompts, APIs and application behavior

Learn how applications call models, how context is assembled, how streaming behaves, and why rate limits, timeouts, retries and provider quotas become operational concerns.

Browse the complete learning path →
03

RAG, embeddings and knowledge systems

Retrieval-Augmented Generation adds ingestion, embeddings, vector databases, ranking, authorization and data freshness to the request path. These are dependencies with their own latency and failure modes.

Learn RAG →
04

Agents, tools and MCP

Agents turn model output into actions. Once an AI system can call tools or use MCP, SREs must reason about loops, state, permissions, downstream services, blast radius and safe containment.

Learn agents → Learn MCP →
05

LLMOps, model serving and GPU capacity

Move into inference servers, GPUs, VRAM, batching, KV cache, routing and autoscaling. This is where familiar capacity engineering returns, but the scarce resources now include context length, tokens and accelerator memory.

LLMOps roadmap → GPU basics →
06

Reliability, observability and SLOs

Measure the experience users actually receive: availability, task completion, latency, TTFT, TPOT, answer quality, retrieval quality, tool success, provider health and cost. Build SLOs around the critical user journey.

Observability guide → SLO concept →
07

Security, incidents and FinOps

Production AI needs trust boundaries, least privilege, prompt/RAG/tool defenses, kill switches, provider failover, incident runbooks and cost controls. Reliability includes safe behavior and sustainable spend.

AI FinOps → AI Security Engineering →
AI SRE SIGNALS

HTTP 200 is only the beginning.

An AI service can be technically reachable and still fail the user. A useful operating model combines traditional service telemetry with AI-specific signals.

Availability

Can the user complete the task, including critical dependencies?

TTFT

How long until the first generated token makes the system feel alive?

TPOT / tokens/sec

How quickly does output arrive once generation begins?

Quality

Is the answer relevant, grounded, safe and useful?

Tool success

Do agent actions and downstream calls actually complete correctly?

Retrieval health

Are the right documents available, fresh and authorized?

Capacity

Are queues, concurrency, GPU memory and provider limits under control?

Cost

What does each successful request or task cost, and is spend drifting?

Production rule: optimize the complete user journey, not the component with the most familiar dashboard.

PRODUCTION ARCHITECTURE

Think in dependencies, not just models.

A modern AI request can cross authentication, an API gateway, prompt assembly, retrieval, vector storage, reranking, model inference, tool calls, policy checks and output validation. Any one of those components can dominate latency or cause the task to fail.

INCIDENT RESPONSE

Triage the failure domain before applying a generic retry.

AI incidents often look similar from the outside but require different mitigations. A slow answer could be queue saturation, provider throttling, oversized context, slow retrieval or a tool dependency. A bad answer could be model regression, stale knowledge, retrieval ranking, prompt changes or policy failure.

SymptomPossible domainFirst SRE question
High TTFTQueue / prefill / providerWhere is the wait time accumulating?
Bad grounded answersRAG / ingestion / rankingDid retrieval quality or source freshness change?
Runaway costRetries / loops / model routingWhat changed in calls, tokens or task completion?
Agent side effectTool / identity / authorizationWhich authority crossed which trust boundary?
WHERE TO GO NEXT

Use AIForSRE at the depth you need.

START FROM ZERO

397 concepts. One production mental model.

No coding prerequisite. Learn one concept at a time, then connect it to the systems an SRE actually operates.