/// BLOG

LOG: BLOG

Ruminations on code, intelligence, and the architectural voids of the modern web.

2026-06-16

Vector Databases - pgvector, Pinecone, Qdrant or Weaviate? How to Choose for RAG and AI Memory

The complete guide to vector databases: what they are and why you need them for RAG, when pgvector in Postgres is enough and when you need a dedicated database, a pgvector vs Pinecone vs Qdrant vs Weaviate vs Milvus comparison (latency, cost, scale), the difference between HNSW and IVFFlat indexes, deployment code in pgvector, hybrid search and metadata filtering, and common migration mistakes.

15 min

LLM Evaluation (Evals) - How to Measure AI Application Quality with LLM-as-a-Judge

2026-06-15

AI & Automation

LLM Evaluation (Evals) - How to Measure AI Application Quality with LLM-as-a-Judge

The complete guide to evaluating LLM applications: how evals differ from tests, how to build a golden dataset from real failures, how LLM-as-a-judge works and why it agrees with humans ~85% of the time, which metrics to measure (faithfulness, hallucination, answer relevancy), how to avoid judge biases, how to wire evals into CI/CD, and which tool to choose - DeepEval, Ragas, Promptfoo or Braintrust.

15 min

MCP (Model Context Protocol) - One Standard to Connect AI to All Your Company's Tools

2026-06-13

AI & Automation

MCP (Model Context Protocol) - One Standard to Connect AI to All Your Company's Tools

The complete guide to MCP: what problem the protocol solves, how the host-client-server architecture works, how it differs from function calling and OpenAPI, how to build your own MCP server in Python and TypeScript, which ready-made servers to plug into your business tools, and how to secure the deployment against tool poisoning.

14 min

AI Agent Memory - Mem0, Zep, Letta or LangMem? How to Choose and Implement Memory Across Sessions

2026-06-13

AI & Automation

AI Agent Memory - Mem0, Zep, Letta or LangMem? How to Choose and Implement Memory Across Sessions

Mem0, Zep, Letta or LangMem - or is a profile table enough? A one-paragraph decision rule, a comparison of licences, benchmark results and prices, cost at 1k and 100k users, memory types, Mem0 code, memory in n8n and GDPR compliance.

17 min

Ollama, vLLM and Self-Hosted LLM - How to Run a Local AI Model in Your Company

2026-06-12

AI & Automation

Ollama, vLLM and Self-Hosted LLM - How to Run a Local AI Model in Your Company

A practical guide to self-hosting language models: when your own model pays off more than the OpenAI API, how to configure Ollama on a laptop and vLLM on a GPU server, which GPU to choose, how to connect applications through the OpenAI-compatible API, and how to deploy monitoring for a private LLM in production.

14 min

Prompt Injection and LLM Application Security - How to Protect Chatbots, RAG and AI Agents

2026-06-11

AI & Automation

Prompt Injection and LLM Application Security - How to Protect Chatbots, RAG and AI Agents

A complete guide to LLM application security: what direct and indirect prompt injection are, what the lethal trifecta is, and how to build 6 layers of defense - from input guardrails through least-privilege tool calling to monitoring - with a tools table, middleware code and a deployment checklist.

15 min

Fine-tuning an LLM on Company Data - When, How and What to Avoid

2026-06-09

AI & Automation

Fine-tuning an LLM on Company Data - When, How and What to Avoid

A practical guide to fine-tuning language models: when to choose FT over RAG or a better prompt, how to prepare JSONL data, QLoRA on an RTX 4090 with Unsloth, and how to measure whether training actually improved the model.

15 min

How to Choose an LLM in 2026 - OpenAI, Anthropic, Google or Open Weights (Prices, Testing, Routing)

2026-06-08

AI & Automation

How to Choose an LLM in 2026 - OpenAI, Anthropic, Google or Open Weights (Prices, Testing, Routing)

The current GPT-6, Claude 5.5, Gemini 3 and open-weight lineup with prices from official price lists, which model fits which task, how to calculate the cost of a whole task, a one-hour selection test and a multi-model router in code.

12 min

TOTAL_RECORDS: 0110

PAGE 10/14

SYSTEM_STABLE