Writing
RSSThoughts on machine learning, complex systems, and the intersection of technology with society.
No writing in this theme yet.
Latest
AG-UI: o contrato aberto entre agentes e interfaces
Como ligar um agente de IA a uma interface sem lock-in. O protocolo AG-UI explicado com um laboratório interativo embutido no post.
Read article →
Certificação Google Cloud PMLE: mapa do exame e da stack de ML
Parte 1 de uma série em cinco artigos sobre o currículo atual da Professional Machine Learning Engineer da Google Cloud.
Certificação Google Cloud PMLE: dados, features e pipelines
Parte 2 da série PMLE: qualidade de dados, BigQuery ML, Feature Store, skew, drift e pipelines de preparação.
Certificação Google Cloud PMLE: treino, avaliação e operacionalização
Parte 3 da série PMLE: notebooks, treino, tuning, avaliação, experimentação e Model Registry.
Certificação Google Cloud PMLE: serving, escala e MLOps
Parte 4 da série PMLE: inferência batch e online, endpoints, pipelines, retraining e monitorização.
Certificação Google Cloud PMLE: Generative AI, agentes e Responsible AI
Parte 5 da série PMLE: Model Garden, Agent Builder, avaliação generativa, fairness, privacy e safety.
Notas acerca de Gestão de Projectos
Os post-mortems leem-se todos da mesma maneira. Um guia de sobrevivencia para gestores de projecto e os que tem de lidar com eles.
The Hidden Trade-offs of AI Development
Everything nobody told you before putting an LLM in production. A deep dive into the hidden trade-offs of AI development.
The Perfect Project Manual (in an Imperfect Galaxy)
The ten deadly sins of project management and their respective penances. A survival guide for dealing with clients and expectations.
A Short History of Artificial Neural Networks
From 17th-century calculus to the deep learning revolution: tracing the key milestones in the development of artificial neural networks.
Transformer-Based Emotion Recognition from Song Lyrics
Fine-tuning encoder transformers for music emotion recognition using Russell's Circumplex Model.
The Eloquent Void
How large language models work: token prediction, pseudo-memory, and the invisible labour behind AI systems.
Dark Boxes, Dark Patterns
The algorithm is not a tool; it is an environment. We do not use it; we inhabit it.
Emergence from Rules
Three interactive simulations. Simple rules. Complex patterns.
2D Cellular Automaton Explorer
Interactive visualization of complex patterns emerging from simple local rules.
Computer Says No
Why most AI projects fail before production: from garbage data to organizational blindspots.
Data Leakage: The Silent Saboteur
What data leakage is, why it silently inflates metrics, and how to prevent it.
Feature Selection: A Simplified Guide
Why feature selection matters, what to remove, and how to improve your model.
Resources for Scientific Writing
A curated list of resources to enhance your scientific writing and research workflow.
Tips for a Great Presentation
Essential tips for crafting a compelling presentation for your thesis defense.