Giuseppe Barbalinardo.

Scientist, Developer, and Inventor.

Ciao,

I am Giuseppe Barbalinardo. I am a senior data scientist at Tonal and part of the Nanotheory Research Group at the University of California, Davis. I am a NSF MolSSI Software Fellows Alumni.

I am interested in the overlapping areas between machine learning, software development and scientific modeling and passionate about Natural Language Processing. During my Ph. D. I researched how to exploit the deepest secrets of math, science and artificial intelligence to simulate heat in materials.

See my resume

Mathematical Models

I can convert ideas into formulas and formulas into code. I obtained a summa cum laude M.Sc. in theoretical physics. I know statistics, probability theory, information theory, and linear algebra, and I use them in my daily research. I have been a teaching assistant for the graduate class of Numerical Methods in the Chemistry Department at UC Davis.

Software Development

I know how to build software. I spent several years of my life as a software engineer in San Francisco, and today, I am the leading developer of kALDo, an open-source code for Anharmonic Lattice Dynamics calculations using Tensorflow. kALDo has been awarded the MOLSSI investment software fellowship, funded by the National Science Foundation.

Machine Learning, AI and NLP

I use AI and Machine Learning models to solve complex problems. I actively participate in meetups and conferences in the Bay Area, and I apply this knowledge to my research. I am one of the founders of ERGO, a dashboard powered by AI and Natural Language Processing algorithms, which aims to provide news with a greater context.

Portfolio

Heat is a mysterious quantity that has puzzled scientists for centuries, and its understanding is today more important than ever. We can simulate heat transport at the nanoscale using numerical models, at its fundamental atomic level. kALDo is a modern Python-based software that implements both the Boltzmann Transport equation and the Quasi-Harmonic Green Kubo method, which runs on GPUs and CPUs using Tensorflow. It is released open-source, for the scientific community to use and develop.

Discover kALDo

In a world where information is accessible by almost everyone and everywhere, the next big challenge is to decouple meaningful data from noise.
ERGO, is the dashboard for validating news. It is a AI powered platform that pulls the latest stories from various media sources to deliver claims with greater context so readers can better understand the landscape of current events.

Discover ERGO

Past projects
I worked as a software developer and I oversaw the iOS team at Grio, in San Francisco. As a consultant, I had the opportunity to help tens of companies to expand and improve their mobile and web platforms, and I have being exposed to many progemming languages.

Download my resume

Publications and Research

Efficient Anharmonic Lattice Dynamics Calculations of Thermal Transport in Crystalline and Disordered Solids

September, 2020 - Journal of Applied Physics

Understanding heat transport in semiconductors and insulators is of fundamental importance because of its technolog-ical impact in electronics and renewable energy harvesting and conversion. Anharmonic Lattice Dynamics provides a powerful framework for the description of heat transport at the nanoscale... Source code

Neural network reproduces vibrational effects in manganese-germanium

June, 2020 - American Institute of Physics, Scilight

The machine learning method has been applied for the first time to a binary compound. Researchers are increasingly using neural networks as a machine learning method to analyze the atomic structure of complex materials. Although the method has been applied to phase transitions in single compounds, its transferability across compositions of binary compounds has never been tested...

Modeling heat transport in crystals and glasses from a unified lattice-dynamical approach

August, 2019 - Nature Communications

We introduce a novel approach to model heat transport in solids, based on the Green-Kubo theory of linear response. It naturally bridges the Boltzmann kinetic approach in crystals and the Allen-Feldman model in glasses, leveraging interatomic force constants and normal-mode linewidths computed at mechanical equilibrium...

A Statistical Physics Model for Demand-Supply Imbalance in Stock Market Prices

June, 2019 - UC Davis

We introduce a novel approach to model heat transport in solids, based on the Green-Kubo theory of linear response. It naturally bridges the Boltzmann kinetic approach in crystals and the Allen-Feldman model in glasses, leveraging interatomic force constants and normal-mode linewidths computed at mechanical equilibrium...

Thermal transport in finite-size van der Waals materials: Modeling and Simulations

July, 2018 - 18th International Conference on Nanotechnology

Alongside with exceptional electronic and optoelectronic properties, two-dimensional van der Waals materials present intriguing heat transport properties, such as ultra-high thermal conductivity. Here we perform molecular simulations to unravel how heat transport in these materials may be tuned upon mechanical strain and chemical transformations...

An introduction to probabilistic programming

Jul, 2018 - Grio Tech Talk, Youtube, SF

In a world of data, where everything is represented by a probabilities, programming languages are providing new kind of native variables, distributions. Starting from an introduction of probability and Bayes theorem, In this video, we show how modern Machine Learning can take advantage of new probabilistic programming...

Compressing sensing using the Least Absolute Shrinkage and Selection Operator

December, 2016 - Presentation at UC Davis

In a world of data, where everything is represented by a probabilities, programming languages are providing new kind of native variables, distributions. Starting from an introduction of probability and Bayes theorem, In this video, we show how modern Machine Learning can take advantage of new probabilistic programming... Source code

An introduction to Quantum Annealing

November, 2016 - UC Davis

In 2011, the Canadian company D-Wave announced D-Wave One, the first commercially available quantum computer based on Rainer, a 128 superconducting flux qubits processor. Since very early age, the D-Wave computer had been used as a heuristic minimizer of the Ising energy functions. In the last few years, many institutions supported the research for the use of this machine as an optimizer...

Server side swift using IBM Kitura

Aug, 2016 - Grio Tech Talk, Youtube, SF

Swift has gained a lot of popularity during recent years. From being a language used to develop only Apple devices, this programming language, has recenlty expanded its applications to many other field. A recent example is Android development. In this video we show how to quickly create a web server usign Swift and IBM Kitura... Source code

Apple TV prototyping with Interface Builder

March, 2016 - Grio Tech Talk, SF

With the acquisition of Next in 1997, a new tool was initiated into the Apple family. Originally known as an enhancement of OpenStep, called NextStep, it caught the attention of the developer community under the name of Interface Builder, as part of the XCode suite. Now about to celebrate its 20th birthday, Interface Builder represents the most powerful IDE to... Source code

Parallel computing with a GPU

May, 2015 - Grio Tech Talk, SF

Graphic Processor Units are becoming more and more important in recent years and are spreading into many different fields, some of which include: computational finance, defense and intelligence, machine learning, fluid dynamics, structural mechanics, electronic biology, physics, chemistry, numerical analysis and security. There are many reasons why a... Source code

A Monte Carlo simulation for predicting the World Cup

July, 2014 - Grio Tech Talk, SF

Several weeks ago I tried to predict who would win the World Cup. I faced this interesting problem I want to share: how can we relate the outcome of the World Cup with the strength of the teams... Source code

Contact