Technical Founder & Full-Stack AI Engineer
I build AI systems that work in production — from architecture to deployment. Specializing in LLM systems, AI agents (OpenAI Agents SDK, MCP), and the training & evaluation data that makes them perform. PhD in NLP. 9 years of applied experience. Founder of Calibrion AI.
What I Do
End-to-end data pipeline design for LLM training and evaluation — HITL workflows, quality frameworks, consensus-based QC, and LLM-jury evaluation.
Production LLM systems including fine-tuning (SFT, DPO, RLHF), evaluation, benchmarking, and LLMOps across leading model providers.
Design and deployment of agentic workflows and multi-agent systems using OpenAI Agents SDK, Claude Agents SDK, Pydantic AI, and MCP.
Deep expertise in NLP — from transformers, RAG, and semantic search to sequence modeling and production ML pipelines.
Translating complex business needs into scalable AI implementations. C-suite advisory on AI strategy, data governance, and product development.
Player-coach engineering leadership — building and leading cross-functional teams, defining technical roadmaps, and driving execution in fast-moving environments.
Selected Work
Founded and led the development of a production-grade annotation platform combining HITL workflows, annotation agents, consensus-based QC, and LLM-jury evaluation — deployed at enterprises including Zalando and Siemens Energy.
Designed and built agentic architectures using OpenAI Agents SDK to dynamically generate components, applications, and user interactions within an existing full-stack framework. Agent orchestration, OpenAI API, and synthetic data generation.
Designed and implemented a RAG system for a DJ school to offer an AI-powered tutor to their thousands of online students.
Designed and delivered AI solution architecture and implementation in NLP, LLMs, and AI agents for one of the world's leading data and AI platforms.
Advised the private foundation of former Starbucks CEO Howard Schultz on AI adoption, delivering production AI implementations aligned with their mission.
Background
2024 – Present
Founder & CEO
Calibrion AI · Berlin
Building a data annotation platform for foundation model training and evaluation data.
2021 – Present
Senior ML/NLP Consultant
Freelance · Remote
AI solution architecture and implementation for global enterprises and mission-driven organizations.
2021 – 2022
ML Lead
Archipelo · San Francisco (Remote)
Hands-on ML team lead, NLP infrastructure, and feature development.
2020 – 2021
NLP Lead
Merantix Labs · Berlin
NLP roadmap, enterprise AI solutions for Volkswagen and Roche.
2018 – 2020
Senior Data Scientist
Omnius · Berlin
Improved document classifier accuracy by 12%. Transfer learning, anomaly detection, and Siamese text mapping for production document AI.
2017 – 2018
Data Scientist
Market Logic Software · Berlin
Built NLU/NLG and a reinforcement-learning dialogue manager for a marketing chatbot deployed for Unilever.
2012 – 2017
PhD in Computer Science
University of Geneva · Switzerland
Specialization in NLP, awarded avec mention très bien (highest distinction). Visiting researcher at Stanford University (Center for the Study of Language and Information) and Uppsala University (Computational Linguistics).
Academic Work
My academic research focused on natural language processing, linguistic annotation, and distributional semantics. My work has accumulated 191 citations and includes papers co-authored with Prof. Joakim Nivre (one of the most cited NLP researchers globally) and collaborators at Google. My most cited work involved building annotated datasets for complex NLP phenomena — an early foundation for the data-centric AI work I continue today.
Let's Work Together
Calibrion AI provides an end-to-end data annotation platform including:
I take on select consulting engagements including:
Contact
I'm based in Berlin and open to remote consulting engagements globally. Feel free to reach out about projects, collaborations, or just to talk AI.