Veronica Tozzo

LinkedIn GitHub adwc1d@r.postjobfree.com 617-***-**** Boston (MA), USA

BIOMEDICAL DATA AND MACHINE LEARNING SCIENTIST

With 6 years of experience at the intersection of biomedical and machine learning research, I have a track record of successfully developing and implementing innovative machine learning models that have led to insights in genetics and clinical research. What drives me is the opportunity to leverage data to drive innovation and make a meaningful impact on patient care. I am excited to continue applying my skills in a dynamic and collaborative environment. PROFESSIONAL EXPERIENCE

PostDoctoral Researcher Mar 2020 – Present

Massachusetts General Hospital, Harvard Medical School Boston (MA), USA

• demonstrated the need of adjusting hemoglobin A1c for red blood cell lifespan on three clinical trials on diabetes

• established the unreliability of current clinical markers for glycemic control and provided translational solutions to limit errors and improve patients diagnosis and monitoring

• successfully built a mechanistic model to determine red blood cell age from clinically available data with applications in early detection of cancer, anemia and improvement of diabetes monitoring

• enhanced the state-of-the-art in deep models on set prediction, effectively leveraging these advanced techniques to accurately predict clinical from single cell flowcytometry data

• refined longitudinal reference ranges in pregnancy and defined intra-pregnancy change as a new marker. Demonstrated that abnormal behavior have a 2x risk of developing complications later in pregnancy compared to previously used markers

• effectively supervisedPhDs candidates and undergraduate interns Research fellow Jan 2020 – Apr 2020

University of Genoa Genova (GE), Italy

• improved computational experimental pipeline for collaboration between local hospital and university by setting up computational resources on local clusters and implementing standardized hypothesis testing approaches Data science consultant Jun 2019 – July 2019

Linear Genova (GE), Italy

• improved company’s methods for signal decomposition and clustering of audio tracks from auditory devices

• provided general tools to employees for future in-company use TECHNICAL SKILLS

Languages : Python, Matlab, R, C++, Java

Libraries : numpy, scikit-learn, pytorch, pandas, wandb, nmle Databases : SQL

Dev Tools : Visual Studio Code, Git, Gitlab, slurm Data types : genetics, clinical, EHR, NGS, GWAS, tabular, time series, medical images, flowcytometry EDUCATION

University of Genoa Genova (GE), Italy

PhD in Computer Science - Thesis title: Generalized temporal network inference Nov 2016 – Jan 2020 University College of London London, UK

Internship on genetic networks integration Sep 2019 – Dec 2018 University of Edinburgh Edinburgh, UK

Internship on genetic and epigenetic data Feb 2016 – Jul 2016 University of Genoa Genova (GE), Italy

Master of Science in Computer Science Sep 2014 – Jul 2016 University of Genoa Genova (GE), Italy

Bachelor of Science in Computer Science Sep 2011 – Jul 2014 PROJECTS

Deep single blood cell models bash, LSF, matlab, python, pytorch, slurm, wandb, git Source Code

• implemented automatic data collections, cleaning and transfer among three servers of over 3 millions data point

(bash, LSF, matlab)

• designed and developed data and code infrastructure for automatic hypothesis testing given data specifics (python, sklearn, pytorch, slurm, wandb)

• improved over state of the art and uncovered new biological insights on blood cells regulation Mechanistic model of red blood cell ageing matlab, SQL, git

• designed a novel approach from biological knowledge and assumptions

• implemented method and optimization in matlab with in-code queries to SQL databases for data collection Regularized graph inference library - ReGAIN python, git Source Code

• gathered and re-implemented the state of the art following scikit-learn guide lines to facilitate experiments pipelines

• designed, derived, implemented and tested novel graph inference methods that are more effective on time series with latent or missing data or that are non-stationary Dictionary learning library - DALILA python, cuda, git Source Code

• achieved a dictionary learning method implementation transparent to regularization choices

• improved on training time by implementing distributed approaches to cross validation and a GPU version of the algorithm (dask, cuda)

SELECTED PUBLICATIONS

• The effects of daily prednisone and tocilizumab on hemoglobin A1c during the treatment of giant cell arteritis Patel*, Tozzo* et al. (2022)

• Set Norm and Equivariant Skip Connections: Putting the Deep in Deep Sets Zhang*, Tozzo* et al. (2022)

• Statistical Models Coupling Allows for Complex Local Multivariate Time Series Analysis Tozzo*, Ciech* et al. (2021)

• A telescope GWAS analysis strategy, based on SNPs-genes-pathways ensamble and on multivariate algorithms, to characterize late onset Alzheimer’s disease

Squillario et al. (2020)

• Secondary somatic mutations in G-protein-related pathways and mutation signatures in uveal melanoma Piaggio et al. (2019)

LANGUAGES

English: Fluent Italian: Native French: Basic

SOFT SKILLS

• excellent communication skills and a high degree of emotional intelligence - when a collaborator’s performance became an issue, I initiated a feedback session to discuss their level of commitment, expectations for contribution, and preferred methodological approach. Together, we developed a more effective division of tasks that accommodated everyone’s needs while still making progress on the project

• supporter of feedback, constructive criticism and transparency on personal knowledge - when leading the discussion I always explicitly ask to be interrupted and corrected when necessary

• great track record of collaborations across and within fields - I have 5 publications as first co-author in machine learning and clinical research

• eager to learn new skills and domains - on a clinical project I learned new biology and methods in under 3 months, this led to a publication in an overall 9 months period

• remarkable time management and organization skills - great at multi-tasking and consistently delivering high quality results, handled 5+ active projects with weekly or biweekly progress reports

• great at keeping code and projects well-documented and efficiently managed - for every project where heavy coding is involved I plan in advance how to structure my code and use github for versioning (see projects above)

Contact this candidate