Rohaan Deb

Data scientist with a physics background, with over 3 years experience building pipelines on complex, high dimensional data in real-world settings. Strong in data preprocessing, feature engineering, anomaly detection and model validation using Python. Curious and self-driven, with a strong interest in applying technical skills to real-world challenges.

PhD | Data Pipelines, Systems & Machine Learning at CERN | University of Trento (2022-11 – Present)

Built and deployed ML pipelines on large, noisy datasets with improved model stability and detection sensitivity. Applied statistical modeling and signal processing techniques to extract patterns and identify anomalies in high dimensional data.

• Built and deployed ML pipelines in Python (scikit-learn, PyTorch) on large, noisy datasets, improving model stability by 20% and detection sensitivity by 25% through structured validation and preprocessing.
• Applied statistical modeling and signal processing techniques to extract meaningful patterns and identify subtle anomalies in high dimensional data, improving signal quality by 30%.
• Worked with high-dimensional sensor and simulation data, applying techniques transferable to computer vision tasks such as pattern extraction, noise reduction, and spatial data analysis.
• Contributed to code reviews and debugging using Git, improving data pipelines and fixing issues to make workflows more stable and efficient.
• Collaborated with physicists, engineers, and analysts to align data processing with real system behavior, and translated complex results into clear, usable insights for non-technical audiences

Data Analyst | Predictive Modeling on High-Dimensional Datasets at University of Bonn, Germany (2021-05 – 2022-09)

Built ML pipelines for high-dimensional datasets including feature scaling, dimensionality reduction, and supervised models. Applied signal processing and statistical methods to improve data quality.

• Built ML pipelines in Python (scikit-learn, PyTorch) including feature scaling, dimensionality reduction (PCA), and supervised models (regression, classification) for high-dimensional datasets.
• Applied signal processing and statistical methods (filtering, fitting, noise modeling) to improve data quality, achieving a 30% improvement in signal-to-noise ratio.
• Implemented model training, cross-validation, and hyperparameter tuning (Grid, Random Search) to evaluate model performance and prevent overfitting.
• Engineered features from structured experimental data and built reusable pipelines for preprocessing, training, and evaluation, enabling consistent and repeatable analysis workflows.

PhD in Physics – CERN | University of Trento (2022-11)

Msc. (H) in Physics – University of Bonn (2020-10 – 2022-09)

Bsc. (H) in Physics – AIAS (2016-06 – 2019-07)