Description

I build AI systems that actually ship — and I help you figure out where AI fits in the first place.

My focus is modern applied AI: RAG (retrieval-augmented generation), LLM orchestration, and agentic workflows, built end-to-end from concept to production. I work AI-first (Claude Code daily), which lets me deliver the output of a small team on my own: backend in Python/FastAPI, frontend in TypeScript/React, all wired together cleanly.

What sets me apart is that I don't just write code — I understand the problem, cut through the AI hype, recommend the right approach, and then build the working system you own. That combination of clear thinking and hands-on engineering comes from my open-source project InterBrain, a RAG-based knowledge system I've been building daily for over a year.

Typical projects I take on:

- RAG / knowledge systems — make your documents queryable, with local/private LLMs where data sovereignty matters (GDPR-friendly)

- AI assistants & agents — LLM-powered features integrated cleanly into real products

- AI-powered web apps — end-to-end, from UX to deployment

- Advising teams on what's actually worth doing with AI, then delivering it

Physicist by training, remote-first, and happy to think a problem through with you before we build.

Industry field of expertise

Languages

German
Native or bilingual
English
Fluent
Portuguese
Basic
French
Basic

Workplace preferences

Can work on-site

Braga (up to 50km)

InterBrain
Founder & Lead Engineer
June 2025 - Today (1 year and 2 months)
crowdfunded
Designed and built in public a Git-based architecture that combines collective and artificial intelligence; 3D interface in React / React Three Fiber, TypeScript throughout. Ongoing playground for LLM orchestration, agentic workflows and RAG with local embedding models via Ollama; full development cycle from concept to deployment.
LLM RAG Git GitHub React.js
Liminal Consulting
Freelance AI Engineer
January 2023 - Today (3 years and 7 months)
Bespoke AI-integration systems for organisations: LLM orchestration, RAG, and transcription/OCR pipelines in production web apps. Shipped a paid multi-user pilot for a school: invoice parsing via multimodal-LLM vision (beyond OCR) with auto-categorisation, LLM-based email triage, chat with voice-message transcription. GDPR-compliant AI voicemail assistant for a medical practice: transcription and intent classification via local open-source models — sensitive data never leaves the practice's own infrastructure.
KIT, Institute IISM
Research Assistant (HiWi)
July 2022 - July 2023 (1 year)
76 Karlsruhe, Germany
Built and shipped an LLM-powered meeting assistant in an interdisciplinary team: Whisper transcription →
speaker diarisation → sentiment/emotion analysis → LLM summarisation. Integrated several open-source ML models behind a unified FastAPI; owned model versioning, audio-pipeline edge cases, and containerised deployment (Docker) end to end.