Featured

Why I Built Purple — And Why It Took 20 Years to Get Here

Why I Built Purple — And Why It Took 20 Years to Get Here

Read article
AI/ML Jun 21, 2026

Ollama-powered purple backend

The era of local LLMs promised tremendous power but often came with crippling limitations: your development environment became tethered to high-end GPU hardware. But what if you didn't have to choose?In this deep dive, we explore how Ollama is revolutionizing model deployment by acting as a standardized, portable runtime layer—think of it as the universal operating system for open-source AI models. We examine the architecture powering Purple: a robust hybrid backend that ensures absolute continuity. Whether you are running demanding, private workloads on your powerful home workstation or working remotely on limited hardware, Ollama allows developers to maintain a single, flawless workflow.Discover how achieving model portability means escaping vendor lock-in and physical constraints simultaneously. We break down the technical advantages of blending local private deployment with seamless cloud access, giving you the operational freedom necessary to build truly universal AI applications.

Read more →
AI/ML Jun 17, 2026

Why I Built Purple — And Why It Took 20 Years to Get Here

I've been thinking about AI since before it was cool to think about AI—studying neural networks, formal logic, and operating system internals at my university in Katowice. At the time, it seemed almost impossible that these fields would ever converge, but they did. The wait, however, felt incredibly long.My early professional experience taught me relentless discipline: building commercial software (around 2001), generating revenue with mobile apps in foreign markets (China, 2007), and iterating through diverse tech stacks at companies like EPAM, Uber, and Pearson.

Read more →
AI/ML Mar 24, 2026

Building my own LLM, because why not?

A while back, I decided to train a language model from scratch—not fine-tuning, not LoRA-adapting—but building it all out: initializing weights to noise and building a custom tokenizer from my own corpus. The goal was deeply technical and slightly irrational: I wanted to understand what happens *inside* these models. Not just at the API level, but during those 2 AM moments when the loss curve starts behaving weirdly.

Read more →
AI/ML Apr 12, 2025

Integrating locally running LLM in Go tools to scan Golang modules

I was writing Go applications since 2016 and I was able to go thru various dependencies management systems including famous at some point DEP and now built-in go modules. While go dependency management came long road in language history it still has one major flaw – all dependencies are downloaded from Internet, mostly from github […]

Read more →
Infra Jan 15, 2025

How I Built a Robust Home Network That Survives ISP Changes

Single device networks Generally, when you sign a contract with an ISP, he will provide you with an all-in-one, cost-effective router that has everything built in. It is a Wi-Fi access point, router with integrated switch, and sometimes, it has an additional built-in battery backup or modem/converter if needed. This setup is cheaper to produce […]

Read more →

Built with technologies I love

This blog runs on open, proven tools — chosen for reliability, not popularity.

Go go.dev The whole engine — HTTP server, templating and a pure-Go SQLite, no CGO. Bazel bazel.build Hermetic, reproducible builds across the whole monorepo. Ollama ollama.com Local LLMs powering the AI writing assistant in the editor.