Daniel F.

AI Automation Engineer — Claude API · MCP Servers · TypeScript Full-Stack

Sole developer behind 12 enterprise apps at Europe's largest motoring association. I build production AI systems with Claude API.

Available for new projects — response within 2 hours
Scroll

About

What I deliver

I build AI agent systems, automated data pipelines, and production-grade TypeScript applications. From intelligent automation that makes decisions autonomously, to MCP servers that let your team query business data through Claude or ChatGPT — I turn your applications into tools that work for you. 15 years of shipping software, now supercharged with AI.

15+ Years
Shipping Software
12 Enterprise Apps
ADAC Nordrhein
Production AI
Claude API · MCP Servers

How I work

From idea to production in weeks, not months

01

Discovery

20-minute call. We define scope, success criteria, and timeline. You get a concrete technical approach — not a vague estimate.

02

Prototype

Working prototype in 5-7 days. Real code, real data, real results. You evaluate before committing to the full scope.

03

Build & Iterate

Daily progress updates, weekly demos via Loom. Clean modular code, deployed continuously. No surprises, no black boxes.

04

Handoff

Documented, deployed, with 2 weeks of post-launch support included. The next developer will thank you.

Interactive Demo

AI Pipeline in Action

A real use case: automated lead generation. The pipeline scrapes thousands of prospects daily from job boards, directories, and news feeds. A local AI model filters 80%+ noise before a single API call is made. Claude then evaluates the survivors, scores their relevance, and sends high-value matches straight to Telegram. What used to take hours of manual research now runs autonomously. Hit Run Pipeline to see it in action.

Scrape
Collect raw records daily
Playwright
0 records

Playwright crawls 15+ sources daily, extracting structured data from job boards, news feeds, and APIs. Headless Chrome handles JavaScript-heavy sites.

← Back
Filter
Eliminate noise locally
Ollama
0 records

Local Ollama model (Mistral) pre-screens every record. 80%+ noise eliminated before any API call — keeping costs near zero.

← Back
Evaluate
Score high-value matches
Claude API
0 records

Claude API analyzes surviving records against custom criteria. Scores relevance, extracts key signals, triggers real-time Telegram alerts for high-value matches.

← Back
Notify
Alert on top matches
Telegram
0 records

Top 3 leads ranked by score are sent as a formatted Telegram message with company details, relevance score, and a direct link. Delivered in real time — no dashboard needed.

← Back

Like what you see? Let's talk about your project.

Message me on Upwork
Selected Work

Projects

01
Screenshot
4-Stage Automated Lead Gen

AI-Powered Data Pipeline

A 4-stage system that scrapes thousands of leads daily, filters noise with local AI, scores with Claude, and delivers top matches to Telegram. Fully autonomous, zero manual review.

Claude API Ollama Node.js MongoDB Playwright
View Details Source Code →
02
Full Production App

Flowly — AI Productivity Platform

AI productivity platform in production. Manage goals, automate recurring AI tasks, upload documents for AI-powered Q&A with source citations. Built and operated solo.

Angular NestJS GraphQL MongoDB Claude API Docker
View Details Source Code →
Screenshot
03
Screenshot
Business Data via AI Chat

Production MCP Servers

Your team asks Claude a question, the AI pulls answers from your business data. Multiple MCP servers with domain-specific tools — search, analytics, reporting — secured with OAuth.

Claude API MCP Protocol TypeScript NestJS OAuth Docker
View Details Source Code →

Get in touch

Have a project in mind?

Send me your project details. I'll reply with an honest technical assessment within 18 hours.

Or message me directly on Upwork