Thoughts and explorations in Computer Vision, Machine Learning, C++, and more.
Notes on Andrew Ng's four AI engineering skills, and the tension between two of them that the article doesn't mention.
The useful skill is not picking the best coding model — it is writing an engineering contract that lets the model, provider, and execution location change without rewriting anything.
A practical, agent-agnostic workflow for turning product intent into versioned specifications, implementation plans, executable validation, and repeatable AI-assisted delivery.
A practical taxonomy of modern C++ idioms, dependency patterns, real-time pipelines, boundary patterns, and production architecture beyond the original GoF catalog.
A practical guide to handling GPU architectures correctly in CMake 3.23+, covering native detection, special values, and production best practices.
A comprehensive guide to setting up, structuring, and managing modern C++ projects with best practices, tools, and workflows.
A deep dive into message brokers vs protocols. When to use RabbitMQ, when to choose MQTT, and how to leverage both in hybrid architectures.
A compact cheatsheet of kubectl commands for everyday cluster inspection, debugging, and troubleshooting.
A comprehensive guide to major detector families including YOLO, RT-DETR, and YOLO-NAS — architectures, trade-offs, and deployment considerations.
A comprehensive guide to using Ninja, the fast build system designed for speed. Setup, commands, and best practices for C++ projects.
Core parallel programming patterns in CUDA, from reductions and scans to cooperative groups and stream parallelism.
Understanding the essential building blocks of CMakeLists.txt files, from basic commands to advanced concepts.
Exploring dynamic_cast, typeid, virtual functions, and enums for determining object types at runtime.