Course Project, Responsible Data Analytics, Delft University of Technology, Delft, The Netherlands
Developed a four-phase analytics pipeline and stacking ensemble (RF, GBM, SVR, MLP) to forecast daily solar irradiance for the Baltic Capitals (RMSE 0.58–0.66 kW·h/m²/day; R²≈0.93) and delivered prescriptive grid-management strategies.
(GitHub Link)
Course Project, Neural Networks for Data Science, Sapienza University of Rome, Rome, Italy
Developed a character-level language model using JAX, trained on the Penn Treebank dataset.
Implemented autoregressive decoding (sampling, greedy, and beam search) entirely in vectorized JAX for efficient, pure-python loops-free generation.
(GitHub Link)
Course Project, Optimization Methods for Data Science, Sapienza University of Rome, Rome, Italy
Built a custom MLP from scratch using NumPy only, and applied L-BFGS-B optimization.
(GitHub Link)
Developed a desktop application for university admissions, handling student enrollment, faculty management, course administration, and grading.
Tech Stack: Python (Tkinter, Pandas, NumPy), SQLite
(GitHub Link)