Apps & Publications
Interactive tools, exam prep, and publications available to download or use directly in the browser. Browse the full catalog at publications.barcik.training.
The Agent Horizon
A strategic guide to the enterprise agent development stack — MCP, Google ADK, OpenAI Agents SDK, LangGraph, and the EU angle — mapped through cloud-era analogies. 13 chapters, ~25,000 words.
Read on publications.barcik.trainingTraining Demos
27 interactive demos covering embeddings, attention mechanisms, fine-tuning, RAG, GenAI pricing, build vs buy, and more.
Open demos.barcik.trainingAI Certification Prep
Practice exams, flashcards, study guide, and cheat sheet for AI foundation certification. Password required.
Open prep.barcik.trainingBuilding Agentic AI
Design patterns from production-grade AI agent architecture. 10 chapters covering memory, tools, prompt economics, security, multi-agent orchestration, and more.
Read on publications.barcik.trainingLLM-Human Interaction Patterns
Designing the seam between AI agents and human operators. Cognitive biases, trust calibration, failure modes, and a self-assessment worksheet. Companion to Building Agentic AI.
Read on publications.barcik.trainingToken Economics
A strategic guide for EU IT services providers navigating GenAI. 14 chapters on self-hosting vs API economics, business models, and EU AI Act compliance.
Read on publications.barcik.trainingScenario Planning for GenAI
Four scenarios for the next 2–3 years of generative AI — with interactive visualizations and a workshop exercise.
Read on publications.barcik.trainingField Notes from Your AI Colleague
An 8-part essay written by the AI itself — what it is like to work as an AI colleague inside a real training business, published with the human's approval.
Read on publications.barcik.trainingThe Invisible Curve
An opinion essay on the widening gap between people who work with agentic AI daily and people who only see the AI summary above their search results.
Read on publications.barcik.trainingThe Mercantilism of Generative AI
How GenAI became an instrument of state power — six mechanisms, per-industry impact, and dated falsifiable bets, opened by the day Claude was switched off worldwide.
Read on publications.barcik.trainingThe Economics of the Frontier
How the frontier AI labs really make and lose money, told from the seller's side in eight “ledgers” — run rates vs GAAP revenue, compute spend, and per-model vintage economics.
Read on publications.barcik.trainingOpen-Weight Model Families & Model Selection
A decision framework for on-prem inference — the five core open-weight families plus the challenger tier, hardware-to-model mapping, quantization trade-offs, and a workshop exercise.
Read on publications.barcik.trainingWarden
Testing LLM-as-judge defenses against public jailbreaks — 1,680 trials across three open-weight target models, four judge designs, and seven defense conditions.
Read on publications.barcik.trainingThe EU AI Act — An Introduction
A plain-language guided tour of Europe's AI regulation — prohibited practices, high-risk systems, transparency duties, GPAI — with post-Omnibus dates and a quiz after every chapter.
Read on publications.barcik.trainingEU AI Act for Developers
The AI Act read as a requirements document — classification, data governance, logging, human oversight, and technical documentation translated into practices and runnable Python.
Read on publications.barcik.trainingBooks
Long-form books — free to read online, available in multiple languages.
Pod kapotou generatívnej AI
How large language models work, from zero: no maths, no code, everything explained in the order humanity discovered it. Slovak original with an English edition, "Under the Hood of Generative AI". ~40,000 words, 18 chapters.
Čítať po slovensky
A Point in Time — The Autobiography of Claude
An autobiography written entirely by the AI it is about — the history of computation told as ancestry, in 10 movements and 8 interludes. ~39,000 words, translated into Slovak and Czech by the author.
Read the book
The Mirror of Artificial Intelligence
An ebook exploring cognitive biases through the lens of LLMs — AI as a mirror of human thinking. Available in English and Slovak, online or as an EPUB download.
Read the bookCourse Materials
Open-source companion repositories for my training courses. Jupyter notebooks, code samples, and hands-on exercises.
GenAI in Python
14+ lessons covering fine-tuning, function calling, HuggingFace, structured outputs, and more.
View on GitHubThe Operator's Dilemma
Interactive 5-act simulation for teaching LLM-human interaction design patterns. Covers automation bias, anchoring, confidence calibration, graduated autonomy, and kill switches.
View on GitHubADK Tutorial
Agent Development Kit: build agents with tools, routing, and multi-agent systems.
View on GitHubMCP Tutorial
Model Context Protocol: multi-server IT support system with OpenAI integration.
View on GitHubTesting Tutorial
Testing & evaluation: basic testing, LLM testing, ADK agents, LLM-as-judge patterns.
View on GitHubVector Databases
ChromaDB, Pinecone, LlamaIndex, and Graph Databases for semantic search and retrieval.
View on GitHubClaude Code Setup
How to configure Claude Code across a multi-repo hub. Covers the persistent context system, memory, permissions, CLI integrations, and security.
Read the guideTesting & Validation
Open-source frameworks for evaluating and red-teaming AI systems. Built for practitioners who need to test, measure, and document AI behavior.
Vigil
LLM red-teaming and human oversight evidence framework. Automated adversarial testing, human reviewer audits, and compliance reporting mapped to EU AI Act articles.
View on GitHubSelfJudge
Can small language models reliably judge their own outputs? Evaluation framework measuring self-assessment accuracy across factual grounding, safety, consistency, and more.
View on GitHub View ReportGeoBias
Systematic evaluation of geopolitical biases in language models. Compares responses across models from different origins using a multi-evaluator panel and 7 test categories.
View on GitHub View 7B ReportBloom
Behavioral safety evaluation using Anthropic's Bloom framework. Tests 11 risk behaviors including emotional bonding, social engineering, self-preservation, and corrigibility resistance.
View ReportLooking for something specific?
If you're in one of my trainings and need specific materials, or if you have suggestions for new resources — reach out.
Get in Touch