2025-present
Machine Learning Engineer | Stealth AI Start-up
Fine-tuned a Flux diffusion model with LoRA for multi-view character generation and built animation-pipeline components for shot reuse, scene consistency and model evaluation.
Raluca-Maria Sandu
AI/ML Researcher and Engineer · Zurich, Switzerland
I am an applied AI/ML engineer with around 20 years of coding experience, strong in communication, collaboration and mentoring . My industry work has mostly focused on visual AI, building systems that classify, understand, generate and reason about images and videos. During my PhD in Biomedical Engineering, I worked in European hospitals on medical imaging research for liver cancer, which led to journal publications and contributions to a commercial product deployed in hospitals. My current technical curiosity is physical AI: how machines can understand, navigate, and interact with our messy, beautiful 3D world. That interest also connects deeply with my life outside work: I love hiking, surfing, skiing, snowboarding, gardening, and birdwatching (taking ornithology classes and trying to play with bird photography).
Experience & Education
2022-2025
Machine Learning Team Lead | Accenture
Led and implemented computer vision and generative AI solutions across 7+ projects, including industrial inspection, synthetic data, multimodal verification and 3D generation.
2017-2022
PhD in Biomedical Engineering | University of Bern
Developed quantitative CT and MRI analysis methods for liver tumour ablation using data from a 100-patient multicentre study; the research contributed to a clinical software product.
2016-2017
Master's Thesis Researcher | Philips Research
Developed segmentation and machine-learning methods for the automated segmentation and classification of skin images.
2014-2016
MSc in Biomedical Engineering | RWTH Aachen University
Specialised in medical imaging, image analysis and computational methods for healthcare.
2010-2014
BSc in Systems Engineering (Minor Applied Computer Science) | Politehnica University of Bucharest
For my BSc thesis project in collaboration with Philips Research I performed digital signal processing (volumetric capnography signals) for respiratory signals in healthy vs patients with respiratory dysfunctions.
Recent work
Featured blog projects
August 18, 2026
Multi-View Image Generation with In-Context LoRA
An honest, work-in-progress attempt at teaching FLUX to keep an object's identity straight across four views
Ever since finishing up the wrinkle segmentation project, I've been circling a bigger question: not just recognizing what's flat in a single image, but understanding an object w...
April 18, 2025
Segmentation of Fine Facial Wrinkles with U-Net
Pixel-wise wrinkle segmentation is still an unsolved task with many applications
Table of Contents Introduction What are fine facial wrinkles and why are these so hard to classify? Model architecture Dataset Preparation and Augmentation Results and Evaluatio...
August 19, 2024
Generative AI Image Tools in Marketing and Design: Opportunities & Challenges
GenAI software & tools in image generation & editing for (mostly) non-technical users.
This post has been written as part of my work at Accenture and can be found on LinkedIn as well. This blog post represents my personal views and subjective experience with these...
June 15, 2024
Fine-tuning with Stable Diffusion XL
Fine-tuning your own images with Stable Diffusion XL - Just another project with my cats
Late to the party as usual, I finally got around fine-tuning my own images with Stable Diffusion XL 1.0 on a free-tier Google Colab using LoRA. Despite many tutorials being avai...