Szymon Pacholski Site

Accelerating Language Models with KV Caching

Published: 2025-12-15

A deep dive into one of the most important optimizations in the LLM world: KV Caching. Learn how it works, why it matters, and what trade-offs it involves.

From POC to Production: How to Avoid Burning Your AI Budget and Common Pitfalls

AI/LLMMLOps

Published: 2025-12-08

Learn how to overcome analysis paralysis in AI adoption, start with small steps, and avoid the 5 most common mistakes that turn AI projects into budget black holes instead of successful implementations.

How to Scale LLM Applications in Production

AI/LLMMLOps

Published: 2025-12-04

Building a ChatGPT prototype takes one evening. Deploying it to thousands of users? That's where the real engineering begins. Learn how to transition from 'works on my machine' to stable production using modern LLM tooling.

RAG or Fine-tuning? How to Choose the Best Approach for Your Product

AI/LLMRAGFine-tuning

Published: 2025-11-24

Building LLM products involves a critical architectural choice: RAG or Fine-tuning? Learn the heuristics to save your budget and time-to-market.

JSON vs TOON: Which Format Works Better with Large Language Models?

AI/LLMArchitecture

Published: 2025-11-17

Comparison of JSON and TOON formats in the context of large language models and token efficiency.

How to Run Your Own LLM Locally in Python — Using Bielik-7B and Hugging Face Transformers

AI/LLMLocal LLMPython

Published: 2025-11-10

Learn how to run the open-source Polish LLM Bielik-7B locally on Apple Silicon using Python and Hugging Face Transformers.

Monitoring Claude AI Agents: Why Standard Tools Don't Work (And What Does)

AI/LLMObservabilityClaude Code

Published: 2025-10-27

Building an AI agent is exciting. Monitoring it in production without the right approach is terrifying. Here's what I learned after 7 days of debugging.

Filter by Category

Accelerating Language Models with KV Caching

From POC to Production: How to Avoid Burning Your AI Budget and Common Pitfalls

How to Scale LLM Applications in Production

RAG or Fine-tuning? How to Choose the Best Approach for Your Product

JSON vs TOON: Which Format Works Better with Large Language Models?

How to Run Your Own LLM Locally in Python — Using Bielik-7B and Hugging Face Transformers

Monitoring Claude AI Agents: Why Standard Tools Don't Work (And What Does)

Data Quality and LLM Hallucinations – Why Language Models 'Make Things Up'

High Computational Costs and Infrastructure in Working with LLMs - How to Reduce Them

How to Secure Data in AI Projects: A Comprehensive Guide

How to Handle Constant LLM Model Changes in Business

LangChain and LangGraph – The Duo That Will Elevate Your AI Agent to the Next Level

Circuit Breaker for LLM with Retry and Backoff – Anthropic API Example (TypeScript)

How to Create Your Own MCP Server

How to Build a Really Good AI Agent - Practical Guide

Analyzing System Prompts: Insights from Claude by Anthropic

RAG and AWS Bedrock - A Brief Introduction

How Model Context Protocol is a Game Changer in AI Integration

Introduction to RAG and Vector Search in NLP

Understanding LlamaIndex: Components, Toolsand Agents for LLM-Powered Agents

Automation - Saving Time and Reducing Errors

LLMs in IT: Beyond Code Generation

Understanding LLM Jailbreaking: Testing AI Safety Boundaries

Hallucinations in AI Language Models: Causes and Solutions

Understanding AI Hallucinations: When Artificial Intelligence Makes Things Up

Understanding and Counting Tokens in LLM Models

Best Practices for Building AI Agents

SmolAgents: Building Custom AI Agents with Hugging Face

Preventing Information Leakage in LLM Applications

Prompt Injection: Understanding the Top Threat to Language Models

What is a Vector Database?

Thread Summarization: Adding Memory to LLM Applications

Why and How to Use Fine-Tuning with OpenAI Models

Prompt Engineering: What Every Developer Should Know

Introduction to LLMs: What Are They and How Do They Work?