curriculum/ai-systems/overview

AI & GPU Engineering•60 Topics

AI & GPU EngineeringModel Serving & vLLMAI-Era Verified

Full-Stack AI & Systems Roadmap

A 28-week curriculum engineering autonomous AI agents, advanced multi-stage RAG, high-throughput model serving (vLLM, TensorRT), quantization, Next.js AI streaming apps, and distributed multi-GPU infrastructure.

Mentor Hub

~28w

Total Duration

13 Structured Stages

Total Topics

Crucial Baseline

Core Stages

Step-by-step Modules

Staff

Specialization

LLMs & GPU Infra

Full-Stack AI & Systems — Timeline Gantt Chart

Click any stage row to jump straight into detailed topics, references, and notes.

Review

Applied

Core

Stage

Wk 1Wk 7Wk 14Wk 21Wk 28

SpanDone

01.AI Systems & Python Performance Fundamentals

wk 1–2

02.Deep Learning & Neural Foundation Mechanics

wk 3–4

03.Transformers & Large Language Model Internals

wk 5–7

04.Prompt Engineering, Structured Outputs & Tool Use

wk 8–9

05.Vector Databases, Embeddings & Semantic Search

wk 10–11

06.Advanced Retrieval-Augmented Generation (RAG)

wk 12–14

07.LLM Agents & Multi-Agent Swarms

wk 15–17

08.Model Fine-Tuning & Parameter-Efficient Adaptation

wk 18–19

09.Full-Stack AI Application Architecture

wk 20–21

10.High-Performance Model Serving & Inference Optimization

wk 22–24

11.Distributed AI Systems & GPU Infrastructure

wk 25–26

12.AI Security, Guardrails & Governance

wk 27

13.MLOps, LLMOps & Production Observability

wk 28

Review Pace

3 wks

Fast pass to fill syntax gaps and refresh basics. Optimize for speed here to build velocity for core modules.

Stages: Python Systems, Security & Guardrails

Applied Pace

8 wks

Hands-on implementation and patterns. Expand what you know into practical production workflows and real-world architectures.

Stages: Prompting, Agents, Full-Stack AI, LLMOps

Core / Deep Pace

17 wks

Highest cognitive load and time investment. Deep mathematical foundations, internals, kernel optimizations, and scale design.

Stages: Deep Learning, Transformers, Vector DBs, Serving & Clusters

How to get the most from this roadmap:

Tick topics as you review them, write quick notes in the built-in note pads for sticky recall, and test yourself out loud on the interview questions. Your progress is saved automatically to your browser.