Qdrant PHP Client
Vector search for PHP 7.2 and 8.*
A PHP client that brings Qdrant vector search to existing PHP applications without forcing a runtime upgrade.
Oleg Patsay
I design systems that scale with the business
AI, Architecture & Complex Systems
I design and evolve software systems that scale with products, teams and business complexity.
12+ years in software engineering · FinTech · GameDev · HealthTech · EdTech · MarTech
12+ years in software engineering
- FinTech
- GameDev
- HealthTech
- EdTech
- MarTech
- Architecture
- Engineering Leadership
- Backend Systems
- AI Engineering
- Legacy Modernization
Selected Work
Obsidian MCP Server
A vault as a tool for any LLM
MCP server that connects an Obsidian vault to Cursor, Claude Desktop and ChatGPT. The model can read, write, search and navigate notes.
How I Approach Engineering
Architecture
Complex systems should remain understandable enough for teams to change them safely.
Responsibility
AI speeds up edits. What lands in the system is the engineer's call.
Change
A system needs to change regularly without losing control of it.
Writing
AI Engineering
Notes on MCP, agents, context engineering, RAG, vector search and integrating AI into real software systems.
Software Engineering
Architecture, complex systems, legacy modernization, delivery and engineering practices that keep software changeable.
Engineering Leadership
Technical strategy, ownership, engineering organizations and leading teams through AI-enabled change.
Featured writing
-
AI Code Review: What to Check in AI-Generated Code
AI has made code generation cheaper, but not understanding. Review shifts from syntax to intent, scope, architecture, business invariants, failure paths, and whether the code should exist at all.
-
Who Is Responsible for AI-Generated Code?
AI changed who produces the text, but not who decides to keep it in the system. Ownership remains with the human who understands the change and accepts the merge.
-
How to Introduce AI into a Legacy System Without Rewriting It
Keep the legacy application as the system of record and add AI beside it as a new layer: adapters, read-only scenarios, shadow mode, and gradual expansion of capabilities.
-
What Is MCP (Model Context Protocol) and How Does It Work?
MCP does not make models smarter. It standardizes how AI applications connect to files, databases, APIs, and other external systems.
About
I'm Oleg Patsay, a software engineer and engineering leader with more than a decade of experience building backend systems, evolving complex software and leading engineering teams.