Filippo Banti
HPC & Biomedical Data Pipelines
HPC & Biomedical Data Pipelines
I am a computer scientist based in Switzerland working at the intersection of high-performance computing, data engineering, and biomedical research.
In my current role, I manage and operate HPC infrastructure, working extensively with SLURM workloads and distributed file systems such as Lustre. This experience gives me a strong understanding of large-scale systems and the challenges of processing and debugging time-series data in complex environments.
Alongside this, I have developed research experience in neuroimaging and machine learning during my time with the SCAN research group at Inselspital Bern, where I worked on deep-learning-based brain MRI segmentation and analysis of large-scale, longitudinal datasets (e.g. EPISURG).
More recently, I have been involved in research on Alzheimer’s disease using ADNI data, developing modular and reproducible pipelines for neuroimaging processing, radiomic feature extraction, and the generation of analysis-ready datasets for downstream statistical and machine learning studies.
I am particularly interested in bridging infrastructure and research by enabling robust, scalable, and reproducible analysis of complex biomedical datasets.
Programming
Python, C/C++, R, SQL, JavaScript, HTML/CSS.HPC & Monitoring
SLURM workload management, Lustre distributed storage, distributed systems, time-series monitoring and logging (i.e. Elasticsearch, Kibana, Grafana).Machine Learning & Data Analysis
Deep Learning, CNNs, RNNs, statistical modeling, clustering, classification, feature engineering, NLP fundamentals.Data Engineering
pandas, polars, JSON/XML processing, data pipelines, dataset transformation and validation.Biomedical Imaging
PyTorch, TensorFlow, PyRadiomics, FreeSurfer, NiBabel, ITK, scikit-image, OpenCV.Development of a modular and extensible pipeline for preprocessing and analysis of longitudinal, multi-center brain MRI datasets (e.g. ADNI), designed to handle heterogeneous data sources and enable flexible configuration of processing steps, including cohort selection, biological filtering, normalization, and deep-learning-based segmentation and parcellation; emphasis is placed on reproducibility and data quality through structured outputs, logging, and validation, enabling the generation of standardized, analysis-ready datasets for statistical analysis and machine learning workflows in neurodegenerative disease research.
View on GitHubAn automated tool for the segmentation and subsequent morphometric analysis of the human brain using T1-weighted (T1w) MRI images. This tool is designed to effectively process brain scans not only from healthy individuals but also from patients with various neurological conditions, providing valuable insights into structural brain changes.
View on GitHubA study analyzing a publicly available dataset on the ongoing conflict between Russia and Ukraine. The project involved conducting various network analytics and exploratory analyses, including a novel, time-efficient implementation of the Louvain method for detecting community modularity.
View on GitHubEmail: filippo.banti.23@gmail.com