Courses¶
Hours are the realistic time for you (skipping intro material), not the platform's marketing number. Status checked September 2026. Rule: never run more than one course at a time; a course is a scaffold for a build task, not a goal in itself.
Phase → week mapping
P1 wk 1–4 · P2 wk 5–8 RAG/evals · P3 wk 9–12 agents/protocols · P4 wk 13–16 production/architecture · P5 wk 17–20 scale/depth · P6 wk 21–24 capstone/interviews
Agentic AI & LLM engineering¶
| Course | Provider | Hours | Cost | When | Why / how to use it |
|---|---|---|---|---|---|
| Hugging Face Agents Course | Hugging Face | 10–15 | Free | P3 · wk 9–10 | Framework-neutral agent fundamentals (smolagents, LangGraph, LlamaIndex units) with a certification project; do the units, skip the ones on frameworks you won't use. |
| Hugging Face MCP Course | Hugging Face + Anthropic | 6–8 | Free | P3 · wk 10 | Build MCP servers/clients end to end; pair with the spec. |
| MCP: Build Rich-Context AI Apps with Anthropic | DeepLearning.AI | 2 | Free | P3 · wk 10 | Short, clean intro to MCP primitives (tools, resources, prompts) — do this first, then the HF course. |
| Introduction to LangGraph | LangChain Academy | 6 | Free | P3 · wk 9 | State, reducers, checkpointers, human-in-the-loop, memory. Written for LangGraph 1.x; the labs map directly to your capstone orchestration. |
| Deep Agents project | LangChain Academy | 3 | Free | P3 · wk 11 | Planning + file-system + sub-agent patterns ("deep agents"); good contrast with plain ReAct loops. |
| Agentic AI | DeepLearning.AI (Andrew Ng) | 6–8 | Free / paid cert | P3 · wk 9 | Reflection, tool use, planning, multi-agent — framework-free, strong on error analysis for agents. |
| Evaluating AI Agents | DeepLearning.AI + Arize | 2 | Free | P3 · wk 12 | Component vs trajectory evals, router/skill evaluation, tracing with Phoenix. |
| DSPy: Build and Optimize Agentic Apps | DeepLearning.AI + Databricks | 1.5 | Free | P2 · wk 8 | Signatures, modules, optimizers; follow with the HF DSPy + GEPA cookbook. |
| Retrieval Augmented Generation (RAG) | DeepLearning.AI | 10–12 | Free / paid cert | P2 · wk 5–6 | Longer RAG course: hybrid search, chunking, reranking, evaluation, production concerns. Skim the basics; do the eval and production modules. |
| Claude Code: A Highly Agentic Coding Assistant | DeepLearning.AI + Anthropic | 2 | Free | P1 · wk 2 | Workflow patterns for AI-assisted development (CLAUDE.md/AGENTS.md, sub-agents, hooks) you'll use all 24 weeks. |
| Anthropic courses + Anthropic Academy | Anthropic | 4–8 | Free | P1 · wk 3 | Prompt engineering, tool use and real-world prompting notebooks; Academy has API/MCP/Agent SDK modules. |
5-Day AI Agents Intensive  |
Kaggle + Google | 8–10 | Free | P3 · wk 11 | Whitepapers + codelabs on agents, tools/MCP, memory, evaluation, and deploying with Google ADK. Good for the ADK comparison. |
| Agentic AI MOOC (Fall 2025) |
UC Berkeley RDI | 12+ (pick lectures) | Free | P5 · wk 17–18 | Lectures by lab/framework leads on agent infra, reasoning, evaluation, safety. Watch 4–5 lectures relevant to your capstone. Predecessor: LLM Agents MOOC (Fall 2024). |
| AI Evals for Engineers & PMs | Hamel Husain & Shreya Shankar (Maven) | 20–30 | Paid | P2 · wk 7–8 (or next cohort) | The definitive evals course: error analysis, judges aligned with humans, CI evals. Oct 10, 2026 cohort reported as the last of 2026; the free evals FAQ covers the core if you skip it. |
| Hugging Face LLM Course | Hugging Face | 10–20 (selected) | Free | P5 · wk 19 | Transformers, fine-tuning, and the newer reasoning/GRPO chapters. Do the fine-tuning chapters before the TRL lab. |
| Post-training of LLMs | DeepLearning.AI | 1.5 | Free | P5 · wk 19 | SFT vs DPO vs online RL with small runnable examples; a compact mental model before fine-tuning. |
Deep learning foundations (depth, optional)¶
| Course | Provider | Hours | Cost | When | Why / how to use it |
|---|---|---|---|---|---|
| Neural Networks: Zero to Hero |
Andrej Karpathy | 15–20 | Free | P5 · wk 17–20 | Build micrograd → makemore → GPT → tokenizer from scratch. The best way to feel attention and tokenization; do at least "Let's build GPT" and "Let's build the GPT Tokenizer". |
| Practical Deep Learning for Coders | fast.ai | 15–20 (selected) | Free | P5 (optional) | Top-down, practical; take only if you want broader DL intuition beyond LLMs. |
| CS336: Language Modeling from Scratch |
Stanford | 30+ (assignments) | Free | Buffer wk 25–26 / after | Heavy: tokenizers, kernels, scaling, data, alignment. For after the 24 weeks unless you go deep on inference. |
Distributed systems, databases & system design¶
| Course | Provider | Hours | Cost | When | Why / how to use it |
|---|---|---|---|---|---|
| Concurrent and Distributed Systems lectures + notes |
Martin Kleppmann (Cambridge) | 8 | Free | P1 · wk 2–4 | The clearest short course on clocks, broadcast, replication, consensus, CRDTs. Watch at 1.5x alongside DDIA 2e. |
| MIT 6.5840 Distributed Systems | MIT | 40+ (labs) | Free | P5 · wk 17–20 | Paper-driven lectures + Go labs (MapReduce, Raft, fault-tolerant KV). Do Lab 3 (Raft) if you want real consensus depth; otherwise read the papers from the schedule. Lectures on YouTube. |
| Gossip Glomers |
Fly.io + Kyle Kingsbury (Maelstrom) | 8–12 | Free | P5 · wk 18 | Distributed-systems challenges (broadcast, g-counter, Kafka-style log, txn KV) against a Jepsen-style checker. Can be done in Python. |
| CMU 15-445/645 Intro to Database Systems | CMU (Andy Pavlo) | 20 (selected lectures) | Free | P1 · wk 3 / P5 | Storage, indexes, concurrency control, recovery. Pick lectures on B+trees, MVCC, logging & recovery. |
| CMU 15-721 Advanced Database Systems | CMU | 10 (selected) | Free | P5 (optional) | Columnar execution, query optimisation; useful if you go deep on analytics/vector engines. |
| System Design in a Hurry | Hello Interview | 4–6 | Free (premium extras) | P1 · wk 1, P6 | Best free interview framework + core concepts; the ML system design counterpart for AI SD. |
Cloud, Java & DSA¶
| Course | Provider | Hours | Cost | When | Why / how to use it |
|---|---|---|---|---|---|
| Develop AI Agents on Azure | Microsoft Learn | 6–8 | Free | P4 · wk 13–14 | Foundry Agent Service, tools, multi-agent with Microsoft Agent Framework. Modules may still use "Azure AI Foundry" naming (renamed Microsoft Foundry in 2026). |
| Azure AI Engineer Associate (AI-102) | Microsoft Learn | 20–30 | Free learning / paid exam | Optional, after wk 16 | Only if a certification helps your org positioning; the learning paths are the useful part. |
| Spring Academy | Broadcom / Spring | 5–10 | Free | P4 · wk 14 | Spring Boot fundamentals refresh; for Spring AI 2.0 use the reference docs and workshop repos in awesome-spring-ai. |
| NeetCode | NeetCode | 60–80 (250 path over 24 wks) | Free / paid | All phases | The DSA track backbone: NeetCode 250 in Python with video explanations. |