LEARN · PRACTICE · THINK TOGETHER
I'm Klyff Toledo, a curious mind. I write, research, and build at the intersection of AI, engineering, and human judgment.
Ph.D. in Artificial Intelligence and Knowledge SystemsDoutor em Inteligência Artificial e Sistemas de Conhecimento
I share what I learn, the questions I’m living with,
and the ideas taking shape along the way.
02 / THE BOOK
My sixth book, The High Leverage Engineer, is ready and free for people who want to turn AI-assisted engineering into work they can trust and explain.
We work through specifications, bounded agent loops, evidence, and delivery—so that speed becomes useful, and responsibility stays visible.
Knowledge is for sharing. Digital editions are free. Printed copies cost only printing and shipping; I charge no royalties for shared knowledge.
downloads
03 / RESEARCH
These are directions I'm exploring, with sources you can examine for yourself.
FEATURED RESEARCH · CURITIBA · 2010–2011
Computer vision in the service of child protection.
A personal scientific record of the challenge: CPU budgets, local and global descriptors, classifiers, convolutional networks and the human decisions behind every review.
Read the scientific record ↗
THE SCALE OF THE CHALLENGETECHNICAL PAPER · OCTOBER 23, 2026
An abstract, explicit equations, three reproducible figures, and a careful separation of attention, cache, and recurrent state—beyond the 270k-token claim.
Read the research note ↗
TECHNICAL PAPER · OCTOBER 2026
Scoped rights graphs and challengeable cross-reference attestations for creative works, on chain and off. Missing evidence stays indeterminate. An exclusive right cannot be spent twice.
Read the paper ↗Published on Zenodo ↗TECHNICAL STUDY · AUDIO & MUSICAL SIMILARITY
Tempo, pitch, synthesis and Euclidean distance. Listen to original examples and inspect matching regions, histograms and local alignment paths.
Read the study and hear the examples ↗ENGINEERING NOTES · 2012
São Paulo, 2012: overloaded emergency-call intake, confirmed locations and medically regulated resource routing. A personal historical record, with selected techniques and Python examples.
Read the study and download Python ↗AGENTIC ENGINEERING
I want to connect clear task boundaries with evidence of accepted outcomes. My starting point is to study model capabilities in context, rather than treat a model name as a guarantee.
SELF-SOVEREIGN IDENTITY
I'm interested in identity systems that let people present verifiable claims without making every interaction depend on a single platform. I start with the standards, then ask what a useful implementation would require.
Personal research framing.
LEARN.KLYFF.ORG / LEARN BY BUILDING
Explore the free training with certificate on learn.klyff.org, then practice with these small exercises inspired by the book.
LEARN.KLYFF.ORG · FREE TRAINING WITH CERTIFICATE
A free course on The High Leverage Engineer, built with agents in hours—including a clone of my own voice. An open source platform for creating complete courses with agents.
Free Training with Certificate ↗FOUNDATIONS · START ANYWHERE
Explore how judgment shapes AI-assisted work. Then define a problem before asking a model to solve it.
Begin with the essays ↓Practice space only. Your text is not sent or saved.
ENGINEERING · BUILD ON THE BASICS
Connect the book's Agentic SDLC ideas to a small, bounded task with a clear definition of done.
Explore the book ↓Practice space only. Your text is not sent or saved.
INQUIRY · FOLLOW YOUR CURIOSITY
Use primary documentation to explore agents or digital identity. Separate a documented capability from your own hypothesis.
Choose a research question ↓Practice space only. Your text is not sent or saved.
Learning has room for small pauses. Ask Chimbler for a reflection prompt. She's a provisional illustrated companion, with a very real enthusiasm for your reading time.
01 / READING & IDEAS
AI moves quickly. I want to make room for the thinking that lasts.
ESSAY · DEC 27, 2025
I explore the human decisions that give AI-assisted work its direction.
PRACTICAL PLAYBOOK · DEC 30, 2025
My playbook for taking responsibility for the result.
ON MY READING DESK
I use reporting alongside primary documentation to find questions worth investigating.
Explore IEEE Spectrum's AI coverage ↗A SMALL DETOUR
And occasionally, a scientist and a very fluffy co-pilot.
04 / VENTURES & PRODUCTS
I care about the journey from technical possibility to a problem worth solving.
Everything starts small, solving one problem at a time.
Building a self-sovereign identity ecosystem for a nation means connecting people, public institutions and businesses around trust. Credentials need to travel between systems; people need control over what they share; and privacy, security and interoperability must work beyond a single organization.
At Luure, we built what many thought could not be built—yet. Our research runs deep, and the engineering is demanding. I’m confident you’ll enjoy hearing and reading about the questions we’ve pursued and the possibilities we’re opening.
Where AI, cybersecurity and blockchain meet—and move forward together.
Explore luure.com.br ↗I connect my engineering work with questions about decision-making and prediction-market infrastructure. Visit Predix's own site for the current product story.
Meet Predix ↗I welcome thoughtful conversations about startups, AI, and products in development. Bring a concrete problem, a difficult assumption, or an idea we can examine together.
Start a conversation ↗A FUTURE LEARNING SPACE
I'm imagining a shared workspace where we can train concepts from the book, try small experiments, and learn from what happens. It's a proposal in development, with no release date yet. A future experience may include a full-body Klyff avatar walking alongside Chimbler; that feature is not available yet.
Coming soon · concept stage01 Start with a question.
02 Make a small experiment.
03 Look at the evidence.
04 Learn. Together. ▍
YOUR NEXT CHAPTER STARTS WITH CURIOSITY.05 / ABOUT KLYFF & FAMILY
I'm Dr. Klyff Harlley Toledo, a Ph.D. in Artificial Intelligence and Knowledge Systems, software engineer, architect, researcher, and author. I bring engineering and business outcomes into the same conversation, and I enjoy making what I learn useful to others.
My wife, Aline Toledo, our son Enzo, and Chimbler are the family in this story. They came with me from Brazil.
Find me on LinkedIn ↗FOR ALINE
Aline Toledo
My wife, my best friend, and my favorite place to rest.
I admire your strength, your courage, and the quiet confidence with which you face life's challenges.
Everything we have built carries your mark. And everything I have become carries a part of you.
MANIFESTO
I want to help shape a future where access to code and the ability to understand, use, and develop AI expand human agency.
Make code, learning resources, and the skills to question systems easier to reach. Meaningful access includes the ability to understand, contribute, and build.
Pair experimentation with evidence, human accountability, and proportionate safeguards. Openness and careful release decisions should strengthen each other.
Keep human dignity, choice, and broad participation at the center. The purpose is to expand what people can learn and achieve, rather than concentrate their dependence.
Direct our curiosity toward scientific discovery, new frontiers, and humanity’s long-term flourishing—on Earth and beyond—with responsibility for the people affected along the way.
Whatever your race, color, sex, gender, physical or mental condition, nationality, or planet of origin: you're welcome. All you need is a willingness to share knowledge and learn alongside others.
Say hello ↗