I am a PhD Researcher transitioning into industry. I specialize in analyzing high-dimensional, chaotic time-series data (Fluid Dynamics) and translating complex physical systems into actionable predictive models.
- 🔭 Currently working on: Forecasting chaotic turbulence using LSTM Neural Networks.
- 🎓 PhD Focus: Wall-bounded Supercritical Fluid Turbulence & Modal Analysis (POD/SPOD) at UPC.
- ⚡ Superpower: processing Terabyte-scale datasets (HDF5) efficiently using High-Performance Computing (HPC) clusters.
- 👯 Looking to collaborate on: Physics-Informed Machine Learning (PINNs) & Time-Series Forecasting.
Languages & Core:
Data Science & AI:
HPC & Tools:
Using Deep Learning to predict chaotic systems.
- Objective: Predict the temporal evolution of turbulent velocity modes.
- Method: Trained a Long Short-Term Memory (LSTM) network on temporal coefficients extracted from POD analysis.
- Result: Successfully forecasted non-linear dynamics with low MSE, demonstrating the bridge between Fluid Dynamics and Sequence Modeling.
- Tech: Python, Keras/TensorFlow, NumPy.
The core computational engine behind my PhD thesis, designed to process TB-scale DNS datasets.
This pipeline is divided into two specialized modules for spatial and spectral feature extraction:
-
Module A: Proper Orthogonal Decomposition (POD)
- The Data Science: Unsupervised Learning / PCA (Principal Component Analysis).
- Function: Performs Singular Value Decomposition (SVD) on massive snapshot matrices to extract dominant spatial features.
- Method: Implemented parallelized POD (PCA) algorithms to reduce dimensionality and identify energy-containing vortices.
-
- The Data Science: Advanced Signal Processing / FFT (Fourier Transforms).
- Function: Combines spectral estimation (Welch’s method) with dimensionality reduction to isolate coherent structures in the frequency domain.
- Method: Implemented parallelized SPOD algorithms to reduce dimensionality and identify energy-containing vortices.
I believe the hardest problems in industry—whether in Tech, Finance, or Engineering—are multivariate, stochastic, and noisy. My background isn't just "coding"; it's rigorous mathematical modeling. I don't just import libraries; I understand the linear algebra and statistics that power them.
