Automatic Segmentation of Battery Cells of Electric Vehicles
Automatic detection and segmentation of battery cells of Electric Vehicles.
Professional Summary
About Me
Dynamic machine learning leader with over 15 years in computer vision and 9 years in deep learning. Recognised for guiding cross-functional teams of 5–10 engineers, delivering mission-critical AI projects ahead of schedule, and achieving measurable performance improvements. Key achievements include completing a drone-vision project 5 months early with a 30% reduction in detection latency, winning a national COVID-19 CT segmentation competition, and leading a COVID-19 diagnostic system that became one of two finalists among 30 companies in the IdeiaGov innovation challenge in Brazil. Published research has been cited over 2200 times on Google Scholar. Expertise spans medical imaging, drone vision, diffusion and radiance field models, and large language model fine-tuning. Adept at driving projects from R&D; to production, mentoring talent, and aligning AI initiatives with business outcomes.
Education
PhD in Machine Learning for Medical Imaging
Radboud University Medical Center
MSc in Computer Science
University of Sao Paulo
BSc (Hons) in Computer Science
National University of San Agustin
Interests
Computer Vision
Deep Learning
Machine Learning
Medical Imaging
Pattern Recognition
Experience
Technical Project Manager of Computer Vision
EmbentionLed an 8–10 member R&D; team to develop an autonomous drone-delivery vision system, achieving a 30% reduction in detection latency through algorithm optimisation. Recovered a project with a 5-month delay risk, delivering on time without scope reduction and improving cross-functional collaboration. Mentored engineers and coordinated with hardware, software, and QA teams to meet regulatory and safety requirements. Reported directly to CEO and COO.Senior Data Scientist
Instech Netherlands B.V.Applied state-of-the-art Deep Learning, from prototype to optimization and deployment. Developed algorithms for classification, regression, and segmentation (2D and 3D) in CT scans. Conducted model evaluation, selection, and validation, ensuring robustness and reliability of deployed models. Communicated findings effectively through reports and presentations to both technical and non-technical audiences.Senior ML Researcher (Consultant)
Deep Design SystemsDeveloped AI pipelines for 3D product generation from text using diffusion models and neural radiance fields (NeRFs). Implemented multi-view neural rendering to create photorealistic 3D environments and improved pipeline efficiency.Advanced Python developer and reviewer
OutlierAIReviewed and developed Python and C++ code generated by a Language Model (LLM) to ensure accuracy and adherence to best practices. Provided expert feedback and mentored junior developers, enhancing their coding skills and knowledge. Improved LLM performance by refining its knowledge base through thorough code evaluations and feedback loops. Implemented quality control processes to ensure the correctness of code and feedback.Mentor of LiveProjects
Manning Publications Co.Description: Guided 10 students to develop Deep Learning projects from data collection, training, testing, and reporting.Teaching assistant
Radboud UniversityDescription: Teaching assistant of the Intelligent Systems in Medical Imaging course for Master students. Supervised a Master student during her graduation project.Team leader & Software Developer
BTC & Motorola MobilityProposed solution for Google acceptance and sanity tests: 80% faster and 200% more precise than previous approaches. Created the Android Automation team and led 5 Software developers. Developed a Macbeth ColorChecker algorithm for the Computer Vision team of Motorola Mobility & Lenovo.Computer Vision Consultant
OrkaPod IncDescription: Reduced the false positive rate of OrkaPod’s face recognition algorithm using OpenCV.Researcher and developer in Computer Vision
PCT Consulting Sao Paulo - New YorkEngineered a real-time video surveillance prototype to detect and track individuals, using Computer Vision.Assistant professor and researcher
San Pablo Catholic UniversityAssistant professor and researcher in Machine Learning and Computer Vision.Developer and Junior Researcher
Cathedra CONCYTEC UNSADeveloped feature extraction algorithms using computer vision, the outcome was published in 3 scientific papers.
Education
PhD in Machine Learning for Medical Imaging
Radboud University Medical CenterThesis on Deep Learning for Localization and Segmentation in Thorax Abdomen CT. Applied Deep Learning for localization and segmentation of organs and abnormalities in CT scans. This research contributed to scientific journal publications.Read ThesisMSc in Computer Science
University of Sao PauloDissertation on Supervised Feature Selection by Ranking to Process Similarity Queries in Medical Imaging. Published scientific papers in conferences.Read DissertationBSc (Hons) in Computer Science
National University of San Agustin
Relevant Projects
Fully Automatic Volume Measurement of the Spleen at CT Using Deep Learning
Automatic spleen segmentation using deep learning. This method reaches radiologist performance.
Deep Learning for Localization and Segmentation in Thorax Abdomen CT
Deep Learning for Localization and Segmentation in Thorax Abdomen CT.
Efficient organ localization using multi-label convolutional neural networks in thorax-abdomen CT scans
Automatic organ localization using Deep Learning.
Recent Publications
(2023).
Transfer learning from a sparsely annotated dataset of 3D medical images.
arXiv, arXiv:2311.05032.
(2023).
Kidney abnormality segmentation in thorax-abdomen CT scans.
arXiv, arXiv:2309.03383.
(2023).
The Liver Tumor Segmentation Benchmark (LiTS).
Medical Image Analysis 84, 102680.
(2020).
Fully Automatic Volume Measurement of the Spleen at CT Using Deep Learning.
Radiology: Artificial Intelligence, 2020;2(4):e190102.
News & awards
Winner of the Brazilian COVID19 detection/segmentation challenge
Government of the State of São Paulo/Brazil ∙
August 2020
I won the automatic COVID19 detection challenge using CT scans. Gabriel was interviewed by the University of São Paulo. This challenge was organized by the Gov of São Paulo.
Languages
100%
Spanish
100%
English
100%
Portuguese
10%
Dutch