Build AI that matters 🌱 is now live!
Build AI that matters is a deck focused on building AI that doesn’t just work in demos - it works in the real world, under pressure, when it matters most.
This isn’t about chasing state-of-the-art or shipping fast. It’s about building AI systems you can actually trust.
Oxford ML Summer School 2025 // Recap 🎓
Spent a week at the Oxford ML Summer School (OxML 2025) - one of the most intense and rewarding deep dives into machine learning I’ve had in a while.
Dependable AI in Airborne Systems ✈️
Integrating AI into safety-critical airborne systems presents profound challenges — this talk addresses the complexities of ensuring reliability, predictability, and compliance with stringent aviation safety standards.
poolside Offsite in Peniche 🥽
Still energized from an extraordinary exchange with poolside’s leadership team, exploring a not-so-distant future where humans and AI agents collaborate seamlessly.
AI x Innovation @ AIHub 🧠
The AI x Innovation team visited the AIHub by Unicorn Factory Lisboa for an inspiring exchange with startups pushing the boundaries of AI.
AI Red Teaming @ Critical Software 🟥
In collaboration with the AIHub by Unicorn Factory Lisboa, we’re hosting a semi-private AI red teaming exercise in our Lisbon office — bringing together enterprises and leading startups to stress-test AI systems in a controlled, ethical, and collaborative setting.
📨 DM me for more details.
Critical Software @ Liga dos Inovadores 💡
The Liga dos Inovadores podcast just released a new episode featuring a conversation about our work at Critical Software and the unique challenges of deploying AI in Defense (OVERSEE) and Space (Karvel).
Safety-Critical AI is now Awesome 🚀
Awesome Safety-Critical AI is a curated collection exploring AI’s role in safety-critical systems — where failure means loss of life, major property damage, or environmental harm.
This isn’t about polishing demos or chasing benchmarks. It’s about anticipating chaos and designing systems that can withstand it.
This isn’t just another awesome list. It’s a manifesto and a call to action!
🌟 Check it out and star it on GitHub!
Command AI Assistant (Bedrock Edition) ⛰️👨💻
A Zsh plugin powered by Amazon Bedrock that analyzes failed commands and provides intelligent, context-aware suggestions instantly.
🙌 Adapted from Sean Smith’s Command AI Assistant powered by Ollama.
👨💻 All code and documentation is available at github.com/JGalego/command-ai-bedrock.
LLM Goblet 🍷 > LLM proxy application powered by AWS Chalice and LiteLLM
LLM proxy application powered by AWS Chalice and LiteLLM.
👨💻 All code and documentation is available at github.com/JGalego/LLM-Goblet.
cl-bedrock 👽⛰️ > Amazon Bedrock meets Lisp
A Common Lisp library for Amazon Bedrock — bringing the power of foundation models to one of programming’s most elegant languages.
👨💻 All code and documentation is available at github.com/JGalego/cl-bedrock.
Serverless LLM Proxy ✨ /֎-Compatible/
An OpenAI-compatible proxy running on AWS Lambda and Amazon API Gateway, powered by LiteLLM — seamless model switching without changing your code.
👨💻 All code and documentation is available at github.com/JGalego/Serverless-LLM-Proxy.
VektorDB 🏹 > Minimal Vector Database for Teaching
A minimalist vector database built for educational purposes — because the best way to understand something is to build it yourself.
👨💻 All code and documentation is available at github.com/JGalego/VektorDB.
Hacking GraphRAG with Amazon Bedrock 🌄
Run GraphRAG pipelines powered by Amazon Bedrock using LiteLLM proxy — combining knowledge graphs with retrieval-augmented generation.
📝 Read the full article on AWS Community.
Make Programs, not Prompts: DSPy Pipelines with Llama 3 on Amazon SageMaker JumpStart
Build ML pipelines with DSPy powered by Meta’s Llama 3 70B Instruct running on Amazon SageMaker — shifting from prompt engineering to programmatic LLM workflows.
📝 Read the full article on AWS Community.
Leaving no language behind with Amazon SageMaker Serverless Inference 🌍💬
Deploy a high-quality translation model at scale using Amazon SageMaker Serverless Inference — bringing machine translation to languages that need it most.
📝 Read the full article on AWS Community.
Build Document Processing Pipelines with Project Lakechain
Create cloud-native, AI-powered document processing pipelines on AWS using Project Lakechain — a framework for building sophisticated document transformation workflows.
📝 Read the full article on AWS Community.
Your idea is my command: multi-agent collaboration for software development 💡🚀
Build your own virtual software company using Amazon Bedrock ⛰️ and ChatDev 👨🏼💻 to develop custom applications through LLM-powered multi-agent collaboration.
📝 Read the full article on AWS Community.
Running LangChain.js Applications on AWS Lambda
Deploy LangChain.js applications powered by Amazon Bedrock on AWS Lambda using function URLs and response streaming.
📝 Read the full article on AWS Community.
Scrape All Things: AI-powered scraping with ScrapeGraphAI 🕷️ and Amazon Bedrock ⛰️
Extract information from documents and websites using natural language prompts — no complex selectors or parsing logic required.
📝 Read the full article on AWS Community.
Bug hunting with Amazon Bedrock and SWE-Agent 👨💻
Build your own software engineering agent using SWE-Agent and Amazon Bedrock to autonomously fix real bugs and issues in GitHub repositories.
📝 Read the full article on AWS Community.
Fighting Hallucinations with LLM Cascades 🍄
Discover how to implement FrugalGPT-style LLM cascades on Amazon Bedrock using LangChain and LangGraph — without breaking the bank.
📝 Read the full article on AWS Community.
👨💻 All code and documentation is available at github.com/JGalego/FrugalBedrock.
Running Small Language Models on AWS Lambda 🤏
Discover an elegant way to deploy small language models (SLMs) or quantized versions of larger models on AWS Lambda using function URLs and response streaming.
📝 Read the full article on AWS Community.
👨💻 All code and documentation is available at github.com/JGalego/SLaMbda.
Mapping embeddings: from meaning to vectors and back
My journey with RAGmap 🗺️🔍 and RAGxplorer 🦙🦺, featuring an accessible introduction to embeddings, vector databases, dimensionality reduction techniques, and advanced retrieval strategies.
📝 Read the full article on AWS Community.
📢 UPDATE: An expanded, interactive version is now available at critical-ai.dev/MappingEmbeddings.
#TGIFun🎈 Building GenAI apps with managed AI services
Reflections on managed AI services and their evolving role in the GenAI ecosystem — with practical examples.
📝 Read the full article on AWS Community.
🧪 The Rise of the LLM OS: From AIOS to MemGPT and beyond
A personal exploration of the fascinating convergence between LLMs and operating systems, with thoughts on how they might collaborate in the near future.
📝 Read the full article on AWS Community.
#TGIFun🎈 YOLambda: Running Serverless YOLO Inference
In this episode of #TGIFun🎈, discover how to deploy any YOLO version 👁️ (YOLOv5-11, and beyond) on AWS Lambda using the AWS SAM CLI.
Deploy LangChain 🦜🔗 applications on AWS with LangServe 🦜️🏓
Discover how to deploy LangChain applications with LangServe on Amazon ECS and AWS Fargate in minutes using AWS Copilot.
📝 Read the full article on AWS Community.
Bedrock JCVD 🕺🥋 on LangChain templates
The Bedrock JCVD 🕺🥋 template has officially joined the LangChain ecosystem!
Bringing 🤗 Text Embeddings Inference to Amazon SageMaker
Just opened PR huggingface/text-embeddings-inference#103 to add SageMaker-compatible images to HF TEI, following the pattern established by huggingface/text-generation-inference#147.
Chatting with Support - Cost Optimization with AWS 💰
A new video on cost optimization featuring Jonas Ferreira — the inaugural episode of our PT-PT series for the AWS Iberia YouTube channel.
A Tour of GenAI 🚀 - There and Back Again
My perspective on the GenAI narrative — no hype, no hubris, no hogwash.
📢 UPDATE: Now live at critical-ai.dev/GenAI!
Time in Machine Learning Engineering ⏳
This article started out as a joke and didn’t wander very far in state space. It is a witty and not-so-rigorous attempt to demonstrate the importance of time in ML projects that will annoy most mathematicians and alienate some physicists. There’s some truth in it… it’s just really hard to find. Enjoy! 😛
📝 Read the full article on Medium
HuggingFace 🤗 on Trainium
A deep dive into training 🤗 Transformers on AWS Trainium — AWS’s custom chip designed for high-performance deep learning.
📢 UPDATE: 🤗 on Trainium is now part of the AWS Iberia AI/ML workshops
QHack Recap - PennyLane, Amazon Braket and Beyond 🚀
Now that QHack 2021 has wrapped up, here’s a comprehensive look back at quantum computing’s most exciting hackathon…
📝 Read the full article on Medium
Defect Detection with Amazon Lookout for Vision 🏭
This article introduces AWS for Industrial with a deep dive into Amazon Lookout for Vision, a service that transforms defect detection into straightforward binary image classification.
📝 Read the full article on Medium
WBME Workshop - Machine Learning for Medicine and Healthcare 👨⚕️
An exploration of machine learning and its transformative impact on healthcare — from the expert systems revolution of the 1980s to today’s AI-driven diagnosis, prognosis, and treatment of SARS‑CoV‑2. Discover how AI is reshaping medicine and what the future might hold.
📝 Full content available on GitHub
For more information about this event, visit 12th WBME - Workshop on Biomedical Engineering