AI Engineer Certification
A complete, zero-to-hireable path to agentic engineering: 15 courses, 126 chapters, each with a runnable lab and a check for understanding. Everything runs offline. The companion code and labs are open on GitHub. Free, no login, no email.
33 seconds · 15 courses, 126 chapters, runnable labs
- Course 08 ch
Foundations: Python and Math for AI
From zero. Real Python you run yourself and only the math you need (vectors, matrices, probability, softmax, gradients). 8 chapters, live.
- Course 110 ch
LLM Fundamentals: Build a Tiny LLM From Scratch
Build a real language model by hand: tokenizer, embeddings, attention, transformer, training loop, generation. The depth vertical. 10 chapters, live.
- Course 28 ch
Prompt Engineering
The first lever and highest-frequency daily skill: few-shot, chain-of-thought, structured output, caching, and eval-driven optimization.
- Course 38 ch
LLM Application Engineering
The literal daily job: APIs and SDKs, streaming, structured output, retries and rate limits, cost control, multi-provider routing, observability.
- Course 48 ch
RAG and Embeddings
The #1 production pattern (70% of teams). Embeddings, chunking, vector search, re-ranking, end-to-end RAG, and the failure modes that bite.
- Course 58 ch
Evaluation and Testing
Evals replace unit tests. Golden datasets, code graders vs LLM-as-judge, regression gates, hallucination scoring. Table-stakes.
- Course 613 ch
Agentic Engineering: How Autonomous Agents Are Built
The center of the certification. An agent is a model in a loop with tools, state and a stopping rule. You build that loop by hand, then add planning, memory, permissions, multi-agent structure, evaluation and a cost model, and learn the protocols that let it plug into anything. Nothing here depends on one vendor product; the applied course after it uses Claude as the worked example.
- Course 78 ch
Building Agents with Claude: API, Agent SDK and Claude Code
The applied unit that follows the vendor-agnostic theory: everything Agentic Engineering taught in the abstract, built for real on one stack. The Messages API tool handshake, the Agent SDK loop, then Claude Code itself as the worked example, subagents, hooks, MCP, skills and context files, a guarded fleet, and a complete workflow. Version-stamped to current model ids and API shapes, so check the docs when a parameter moves.
- Course 810 ch
AI Security: Models and Agents
The OWASP LLM Top 10 and prompt-injection offense and defense, then the agent attack surface: the lethal trifecta, injection through tool results, MCP tool poisoning, the confused deputy, and red-teaming your own agent. Ada's differentiator.
- Course 98 ch
AI Engineering in Production
Ship and operate: serving, quantization (AWQ/GGUF), vLLM/KV cache, eval-gated CI/CD, cost and drift monitoring, LLMOps.
- Course 108 ch
Training and Fine-tuning
Shape a base model: datasets, loss curves, SFT and LoRA/QLoRA, DPO, and the judgment of when to fine-tune vs prompt vs RAG.
- Course 118 ch
Transformers Deep Dive
Explain the internals cold: positional encodings (RoPE), attention variants, BPE, KV cache, scaling laws. The T-shape depth vertical.
- Course 126 ch
Multimodal AI
Vision-language, image generation, speech (STT/TTS), and multimodal RAG. The converging edge.
- Course 139 ch
Interview Prep and System Design
Ace the interview: the real loop, from-scratch coding, LLM system design, take-home patterns, behavioral, and question banks.
- Course 146 ch
Capstone Projects
What gets you hired: 4 deployed, evaluated builds. RAG assistant, an autonomous agent with evals and permissions and a cost report, reusable eval pipeline, self-red-teamed app.
Take the final exam and get your certificate.