Let's talk it through

People often write to me to think through a technical decision. This page explains what I can actually help with and how it usually goes — so you know what to expect before writing.

What I can help with

.NET architecture

Minimal API, layer boundaries, MongoDB, gRPC, streaming. Where to simplify and where structure actually pays off.

AI in production

RAG, hybrid search, MCP, function calling, agent orchestration: model choice, cost, guardrails and fallbacks.

Vibe coding and agent workflows

How to set up hooks, skills, quality gates and multi-model review without turning the assistant into a tech-debt generator.

Developer career paths

What to learn, what to skip, and how a full-stack profile evolves after twenty years on the job.

How it usually goes

You send the context

A couple of lines on what you're building, where you're stuck and what constraints you have. The more concrete, the better.

We talk it through

Over email or on a call, depending on how involved it is. We look at the real problem, not the simplified version.

You leave with something concrete

We analyze it together and implement it using the technologies that best fit the problem — zero fluff, you walk away with something that works.

My strength

I don't have to sell anything, and that is exactly my strength — you write to me to discuss and exchange ideas, I have nothing to sell you.

Write to me

I answer personally. A couple of lines of context are enough to tell whether it is worth talking.

Projects I've built

OCR flow: a terminal with incoming documents, the LLM extraction engine and the record created in the CRM

AI OCR for business documents

A pipeline that reads invoices, contracts and orders from PDFs and scans: layout analysis + LLMs to extract fields, business rules to validate them, and typed APIs to push everything into the CRM or ERP.

.NET · OpenAI · Anthropic · Ollama · MongoDB

Abstract illustration: an agent at the centre connected to knowledge base, data, task and dashboard panels

Assistant agent on company data

A private conversational assistant: it answers over the internal knowledge base with RAG and source citations, runs tools via MCP against CRM and ERP, and generates reports and charts on demand.

.NET · Next.js · MCP · OpenAI · Anthropic · Ollama · Qdrant · MongoDB

Chalkboard sketch: records, email, documents and orders converging into a RAG agent that updates the customer card

AI agent integrated into the CRM

An agent with RAG over different sources — records, email, documents, ERP — that unifies data in the context of the customer card: precise answers and records updated directly through APIs.

.NET · MCP · OpenAI · Anthropic · Ollama · Postgres · Qdrant