Ioan-Mihail Dinu Email & Phone Number

New York, Ny, Us

Iasi Metropolitan Area

Bucharest Metropolitan Area

Tenured Research Position involving Machine Learning based optimisation of ongoing physics analyses, statistical analysis and software development.

Măgurele, Ilfov, Romania

Machine Learning on FPGA: - Developed a Convolutional Neural Network inspired by computer vision for muon track identification and a convolutional regression model for candidate track prediction using TensorFlow. - Investigated model quantization options and optimization strategies to minimize FPGA inference time.Unsupervised Machine Learning for Anomaly Detection: - Implemented Autoencoder-based ensemble methods for model-independent multijet BSM searches using python. - Successfully enhanced out-of-distribution detection performance and mitigated model bias by combining Normalizing Flow density estimation and regularized autoencoder reconstruction error.Data Pipeline Automation - Designed and implemented sophisticated automated data pipelines, encompassing Monte Carlo simulation, detector response modeling, and neural network input and target generation.

Geneva, Switzerland

ATLAS Collaboration Member and Qualified Author since 2021EGAMA Combined Performance Work Group - Heavy Flavour Electrons Qualification Task:- Seamlessly integrated heavy-flavour electron selection for ttbar events into the established C++ analysis framework, resolved pre-existing memory leaks, and optimized code for efficient grid system execution.- Executed precise data transformations and thorough data analysis to validate results, ensuring the integrity of the research findings.- Authored comprehensive documentation and crafted engaging visualisations for the internal Research Note, effectively communicating key insights and outcomes.

New York, New York, United States

- Developed and deployed advanced Machine Learning ensemble models, achieving accurate end-of-month revenue predictions within a 15% error margin.- Single-handedly designed and implemented a cutting-edge chatbot Proof-of-Concept, utilizing Large Language Models, Vector Databases, and an intuitive Graphical User Interface.- Streamlined the productionalization of multiple machine learning models, generating valuable insights on B2B sales opportunities and assessing churn risk effectively.- Leveraged Natural Language Processing sentiment analysis techniques to accurately measure client acquisition likelihood, driving data-driven decision-making.

New York, New York, United States

- Expertly implemented a Google Analytics 4 Data Pipeline, integrating data into an S3 Data Lake and a PostgreSQL Database to power a dynamic Tableau Dashboard for comprehensive data visualization- Leveraged langchain to create a chatbot interface to a knowledge based comprised of hundreds of the organisation's documents- Enhanced Marketing Attribution through clever use of Markov chain Monte Carlo

Doctor Of Philosophy - Phd, Elementary Particle Physics

Activities and Societies: CERN ATLAS Group ResearchCo-tutorship with University of Bucharest. Work on PhD thesis centred around the use.

Doctor Of Philosophy - Phd, Computational Physics, Summa Cum Laude

Activities and Societies: CERN ATLAS Group ResearchPhD in co-tutorship between the university of Bucharest (RO) and Université Clermont.

Master Of Science - Ms, Theoretical And Mathematical Physics

Dissertation Thesis: Machine Learning Applications in High Energy Physics

Bachelor'S Degree, Physics

Diploma Thesis: Study of top-anittop quark pair production using Monte Carlo Simulations generated with PYTHIA