AI in your processes
I find the tasks where AI pays off and take them from idea to a working pipeline in production.
>_Fractional FDE · AI implementation
I implement AI in businesses so that it works in production, not just in a demo.
25 years in software development, 11 years in HR tech. Everything I talk about, I first build and run in my own company and for clients.
25
years in software development
11
years in HR tech
What I do
I work as a Fractional FDE (forward deployed engineer): I join your team part-time and take AI use cases all the way to production.
MCP and agents
I connect AI assistants to your products and company data through MCP, so the agent does the work itself.
Evals and quality
I build checks that catch model mistakes before your customers see them.
Hiring and HR tech
My first niche and source of case studies: hiring automation, compensation analytics, platforms for candidates.
Projects
- P-01 NEWHR client IT hiring and market research. AI development and implementation on contract. newhr.org
- P-02 Hello New Job! client A community and courses for job seekers: lessons, mock interviews, AI resume reviews. AI development and implementation on contract. hellonewjob.org
- P-03 Geekjob client A hiring platform where candidates and employers interact through AI agents over the MCP protocol. AI development and implementation on contract. geekjob.ai
-
P-04 TalentPro Vision client Development of an HRM system with an AI assistant.
Approach
A few rules I follow when implementing AI and judging the result.
-
01
Production, not demos
A pretty demo is worth nothing. The only thing that counts is what runs every day on real data.
-
02
Show, don't tell
First a working product, then the presentation.
-
03
Measure quality
Every AI use case has tests and a metric. Without them you cannot tell whether it got better or worse.
-
04
Start simple
A prompt first, then RAG, then fine-tuning. Complexity is added only when the simple approach hits its ceiling.
AI without the magic. I only talk about what actually works.
Articles
I write about things I have built and shipped myself: how they work, where the limits are, and when they are worth using.
- MCP: how to give AI access to your products and data What the Model Context Protocol is, how it works after the 2026-07-28 revision, how to write your own MCP server, protect it with authorization and keep the model from doing too much.
- RAG from scratch: how to make an LLM answer from your documents What RAG is, when you need it, how embeddings, chunking, hybrid search and reranking work, and how to build a first working version on MongoDB step by step.
- Evals: how to check that your AI works beyond the demo What evals are, what they consist of, how to grade answers with code, an LLM judge and a human, how to build your own dataset and which tools exist for this in 2026.
Have a task for AI? Let's talk.
Tell me about your task using the form below. I will reply by email or in your messenger.