Building AI Agents

Free

A curated, sequenced path from language-model foundations to production-grade agents. This is not a link dump. ~59 hand-picked resources are re-sequenced into the six stages that mirror how an engineer actually becomes productive with agents: understand the substrate, learn the control patterns, wire in tools and memory, then measure and operate. 7 modules · 30+ lessons · 3 quizzes · ~90 hours of curated study · Certificate on completion. OUTCOMES OVER CONSUMPTION — the success metric for this path is a shipped agent that passes an eval suite. Every stage names a concrete build target, so progress is measured by working software, not hours logged. Module 1 — Foundations: the language model, tokens, context window, and failure modes. Module 2 — Agent Patterns: the reason-act-observe loop, workflows vs agents, ReAct from scratch. Module 3 — Tools & Memory: MCP, retrieval-augmented generation, vector memory, browser/computer use. Module 4 — Evaluation & Operations: eval suites, accuracy, LLMOps, multi-agent orchestration. Module 5 — Foundational Papers: the primary sources behind modern agent design. Module 6 — Build Habit & Stay Current: hands-on repos and a signal-over-noise cadence. Best for engineers and bootcamp interns who want the fastest credible route to a shipped, evaluated agent.

Instructor: IntelliForge Research

31 lessons

Free

Course content

How to Read This Path

•Start Here — How This Path Works(10m)
•The One Metric — Outcomes Over Consumption(8m)

Stage 01 · Foundations — the substrate under every agent

2 recordings

•Stage Overview & Build Target(10m)
•LLM Introduction — What Is a Language Model, ReallyRecording available(60m)
•LLMs from Scratch — Build the Mechanism Bottom-UpRecording available(60m)
•Go Deeper — Books & Courses(20m)

Stage 02 · Agent Patterns — control, not magic

3 recordings

•Stage Overview & Build Target(10m)
•Building Effective Agents — The Reference Guide (Anthropic)(30m)
•Building Effective Agents — The TalkRecording available(45m)
•Building an Agent from ScratchRecording available(60m)
•Agentic AI Overview — StanfordRecording available(75m)
•Vendor Playbooks & Pattern Catalogs(20m)

Stage 03 · Tools & Memory — giving the agent hands and recall

1 recording

•Stage Overview & Build Target(10m)
•MCP — The Tool-Integration Standard (Anthropic)(20m)
•Building Agents with MCPRecording available(50m)
•Retrieval & Vector Memory (RAG)(20m)
•Agent Memory — Short- and Long-Term Recall(15m)
•Acting on Real Interfaces — Browser & Computer Use(15m)

Stage 04 · Evaluation & Operations — the demo-to-production gap

1 recording

•Stage Overview & Build Target(10m)
•Evaluating AI Agents — The Core Discipline(20m)
•Building & Evaluating AgentsRecording available(60m)
•Operations — Running Agents in Production (LLMOps)(15m)
•Multi-Agent Orchestration — Scope Discipline(12m)

Stage 05 · Foundational Papers — read the primary sources

•How to Read the Papers (Reading Order)(8m)
•Chain-of-Thought — Reasoning in Steps(20m)
•ReAct — Reasoning + Acting(20m)
•RAG Survey — Retrieval-Augmented Generation(25m)
•Toolformer, Tree of Thoughts, Reflexion & Generative Agents(25m)

Stage 06 · Build Habit & Stay Current

•Hands-On Repositories — Keep Building(15m)
•Newsletters — Cadence, Not Noise(10m)
•Capstone — Ship an Evaluated Agent (Priority Matrix)(30m)