Dario Bertazioli

Machine Learning:
, Pytorch, TensorFlow (Keras), Sklearn, Scipy, Statsmodels
NLTK, Gensim, OpenCV.
Data Viz:
Seaborn, Matplotlib, Plotly (Dash), Folium (Leaflet), Tableau, GGPlot, Gnuplot.
Data Science: Numpy, Pandas, bs4, Scrapy, Ray.
Others: R, / (Bash, ROOT(Cern) entry level).
Reporting: $\LaTeX$, Rmd, Md, and HTML.

Physics Background:
Mathematical methods for Physics, Linear Algebra/Geometry, Complex Analysis, Statistics
Quantum Mechanics, Electromagnetism, Structure of Matter, General Relativity.
Data Science Background:
ML & DL, Decision Models (Optimization), Data Analysis, Data Semantics (Ontologies/LinkedData/KG), Text Mining (NLP), Time Series (Linear, Arima, UCM, KF, ML), Computer Vision.

Many years of volunteering left me with
Planning, Communication, Listening capabilities,
and Flexibility.

A Data Management project:
a scalable pipeline for storing meetup.com rsvp streaming data in Arango-db,
analyzing the social network, extracting statistics and visualizing (on a cool folium map!).

The sequel of the "ArMeetup" project with a DataSemantics focus:
a pipeline consisting of online json-ld Ontology population, data enrichment via various APIs, NLP processing for topic extraction, and conversion in RDF graphs for allowing further queries.

Optimization, Hybrid algorithms implementation:
boosting Ant-Colony with Reinforcement Learning (Q-learning),
scaling Genetic Algorithm with k-means clustering.
Benchmarked on the Traveling Salesman Problem (TSP)

Predicting energy supply function for the electricity Italian market:
Functional time-series forecasting with both a statistical (Reduced Rank Regression)
and a machine learning (LSTM based) approach.

A Computer Vision project:
voice recognition (training a VGGVOX-like CNN on custom data), face recognition (finetuning a pretrained CNN), and image retrival via cosine similarity on a KD Tree. Repo link coming soon!

Time Series Forecasting project:
predicting the tidal levels in Venice,
exploiting both linear (ARIMA, UCM) and non-linear (LSTM, GRU) models.