O'car Campos

Self-hosted storage is almost always more cost-effective for large-scale, persistent needs, specially if there is a focus on data privacy and security. For GPU compute (Deepseek), cloud is often cheaper unless you maintain high utilization (>20–30%) so there is to be mindful of the utilization rate and what the goals are before investing. After some research, a hybrid approach—self-hosted storage, cloud GPUs for burst analysis—is optimal for most organizations but all depends on the use case and goals. See detailed cost, risk, and operational breakdowns below for informed decision-making given your own scenario…

I have been asked to design a machine capable of running Deepseek locally for a company that is interested in leveraging AI to improve their processes. The company is strict about data governance, security, and—importantly—budget…

Last year, I started using my indoor grow tent to plant edible vegetables and, at the same time, implement a control system using IoT. For this, I chose Arduino for its ease of programming and learning in prototyping. The system works very well, but little by little I have been iterating components and improving the system to reach something that gives me peace of mind. Warning: I'm just getting started…