The AI Codex SkillUp Machine Learning · Applied AI

This educational platform is designed for everyone who, faced with the complexity of these topics, decides not to learn. Let's make it simple.

dedication
Machine Learning & Applied AI

How do you understand with technical rigor?

An editorial platform for students, self-learners, and teams that need to understand foundations, architecture, and operations without empty explanations.

Built for learning

A learning structure built to enter, deepen, and connect theory with real systems.

01

For students

A clear path through foundations, transformers, evaluation, agents, and systems without arbitrary jumps.

02

For self-learners

Visual explanations and labs that build technical intuition without relying on hype or empty marketing.

03

For technical and business profiles

Useful material for making better decisions about cost, quality, risk, and AI system operations.

Program

An editorial route for learning Machine Learning and applied AI in cumulative layers.

The six codices are ordered so each stage adds a new layer of understanding: foundations, memory, transformers, ecosystems, agents, and operations.

Labs

Labs that turn concepts into observable mechanisms.

The labs are designed so students and readers can look inside models, training dynamics, preferences, and interactive systems.

01 Internal mechanics

Internal LLM simulator

A concrete view into tokenization, embeddings, layers, and step-by-step generation.

02 Continual learning

Nested learning

Layered learning, retention, and refinement for explaining continuity and forgetting.

03 Reinforcement learning

RL Playground

Exploration, reward, and policy behavior observed inside an interactive simulation.

04 Applied alignment

RLHF Explainer

Preferences, reward models, PPO, and the risks of mis-specified optimization.

05 2026 systems

Inference-time compute lab

Reasoning budgets, verifiers, and cost-latency-quality trade-offs in 2026 systems.

06 Multi-agent systems

Multi-agent fraud simulator

Autonomous agents collaborating to detect bank fraud using LangChain, LangGraph, and LangSmith.

07 Vector representation

Embedding Projector

Visualize word embeddings in 3D. Explore semantic relationships with PCA, t-SNE and UMAP.

Francisco Vidal Ureta

Principal Data Scientist & AI Strategist

Academic background

MSc Economics MSc Artificial Intelligence Diploma in Finance Diploma in Big Data & Data Science Applied Data Science Program U. Michigan Universidad Adolfo Ibáñez University & executive teaching

Experience & Consulting

CORFOPuerto San AntonioBanco BCIEuroAmericaCODELCOG&N
AI StrategyMLOpsDecision IntelligenceUniversity Teaching
LinkedIn