Description

I have a PhD in Materials Science, and use my domain knowledge and AI/ML tools to find new solutions which helps inustries to make smarter decisions and faster. I worked with large imaging datasets and developed computer vision pipeliens with physical constraints included in the data and ML pipelines.

Industry field of expertise

Languages

English
Fluent
German
Conversational
Portuguese
Basic

Workplace preferences

Can work on-site

Zurich (up to 50km), Lausanne (up to 50km), Basel (up to 50km), Geneva (up to 50km)

FreeLance
Data and Image Analyst
CONSULTING AND AUDITS
July 2023 - Today (2 years and 11 months)
Zürich, Switzerland
• ● Reduced manual analysis in scientific imaging by developing GPU-accelerated GAN-, UNet-, and Vision Transformer-based pipelines for segmentation, and automated feature extraction from low-SNR datasets.
• ● Accelerated technical literature review and scientific knowledge discovery by building RAG- and agent-based AI assistants using LangGraph, OpenAI embeddings, Gemini, Ollama, and vector-search pipelines, enabling analytical reviews for keywords suitable to R&D teams for new production.
• ● Built production-style ML systems for short-term energy forecasting across 10 European countries using Boosting, Optuna, and MLflow, enabling automated comparison of forecasting models on 800K+ hours of time series data.
• ● Developed a Bayesian optimization workflow using Gaussian Process surrogate models to explore physical properties and optimize them in complex materials.
• ● Addressed limited labelled data and class imbalance by developing physics-guided synthetic data generation pipelines combining simulations, AI-assisted segmentation workflows, enabling more reliable training.
Computer Vision Agnetic AI Machine learning time series data
ScopeM Service Centre, ETH
Technology Consultant
February 2019 - July 2023 (4 years and 5 months)
Zürich, Switzerland
• ● Failure analysis of materials under load using microscopy/spectroscopy, delivery to 15+ clients.
• ● Improved interpretation of experimental microscopy data by integrating Python-based image simulation, spectroscopy analysis, and computational modelling into unified analysis workflows.
• ● Extracted week chemical signals from spectroscopy data using signal processing methods (FFT, PCA denoising, clustering) for two clients.
• ● Applied forecasting and time-series decomposition to analyse elemental concentration of gases, revealing significant patterns and trends.
• ● Delivered technical reports and presentations to academic and industrial partners, translating complex analytical results into actionable technical feedback.
• ● Trained 20+ users on microscopy, image simulation and signal analysis of spectroscopy data.
CICECO Institute of Materials, University of Aveiro
Postdoctoral Fellow
March 2017 - December 2019 (2 years and 9 months)
Aveiro, Portugal
• ● Developed regression and predictive models on experimental and literature datasets to forecast material properties.
• ● Coordinated research collaborations across 3 countries, as microscopy lead in an EU-funded project.