I am a Software Engineer at Apple Inc. in the Autonomous Systems group. Before joining Apple, I was a Postdoctoral Researcher at the Lawrence Berkeley National Laboratory, where I developed a highly-scalable hyperparameter optimization software, HYPPO, that uses surrogate modeling, uncertainty quantification and asynchronous nested parallelism to provide robust and reliable machine learning models for scientific applications. In a previous project at Berkeley Lab, I also worked on deep learning applications for Distributed Acoustic Sensing data from Dark Fiber Testbed for groundwater mapping and broadband seismic event detection. My current areas of interest are machine learning, sensor network technology, and data scalability in high-performance computing environments. Prior to Berkeley Lab, I was a Research Scientist at the University of California, Berkeley where I specialized in magnetometer networks for coherent searches of magnetic transient events. One of the networks I worked on is part of an international effort to detect inhomogeneous dark matter and consists of over a dozen magnetically shielded high-sensitivity atomic magnetometers, each of which is synchronized to the GPS timing. Another network I worked on consisted of a distributed detector of triple-axis fluxgate magnetometers aimed to study the magnetic pulse of cities, that work was done collaboratively with the Center of Urban Science and Progress in New York City. I completed my Ph.D. in Astrophysics at the University of New South Wales in Australia, where I analyzed quasar absorption line data to investigate whether the values of fundamental constants of nature might have been different in the early Universe.
Under the guidance of Prof. John Webb
Abstract: Quasar absorption lines are used extensively in astrophysics to place constraints on cosmological models. In this thesis, we focus on the measurement of two important parameters in cosmology, the electromagnetic coupling constant, or fine-structure constant, α≡e2/(4πεℏc), and the primordial deuterium-to-hydrogen ratio, D/H. Any cosmological variation of α will cause its value to be different in the early stage of the Universe, therefore impacting the production of the primordial light elements during Big Bang Nucleosynthesis. We provide updated ∆α/α measurements from 280 absorption systems previously published in the literature using the non-linear least-square Voigt Profile fitting program VPFIT10. We also investigate the impact of long-range wavelength-scale distortions on those measurements. We found that long-range distortions are unlikely to explain the 4.1σ evidence for an α dipole as reported in the literature even though they do impact on the α-dipole significance. The above work led us to examine the kinematics of each absorption system in our sample. We report a correlation between the complexity of the velocity structure of Damped Lyman-α systems and the apparent position of the Lyman limit break in quasar spectra. We develop a new technique, based on this correlation, to identify suitable Damped Lyman-α systems for D/H measurements.
Under the guidance of Prof. Sebastián López & Dr. Pasquier Noterdaeme
Abstract: According to the current theory of standard Big Bang nucleosynthesis, a measurement of the primordial deuterium-to-hydrogen ratio (D/H) can provide a strong constraint on the baryonic density in the universe. Indeed, D/H depends sensitively on Ωb, and can be thus considered as a “baryometer” of choice. However, because deuterium is easily destroyed inside stars, it is necessary to look for pristine gas (i.e. unprocessed by stars) to make a correct measurement of the primordial D/H ratio. To date, the only way to detect deuterium at high redshift (where the chance of finding pristine gas is higher) is through the absorption lines it imprints in the spectra of background quasars. However, this is observationally a challenging task and very few measurements have been performed so far. In this thesis, I present a systematic search for deuterium absorption lines in a large sample of high-resolution, high signal-to-noise ratio, archival spectra of quasars observed with the Ultraviolet and Visual Echelle Spectrograph mounted on the Very Large Telescope. New deuterium features have been performed and will be analyzed. I will also introduce a new python routine allowing the automatic selection and analysis of suitable candidates for the stacking method.
Under the guidance of Dr. Pierre-Antoine Delsart
pipcommand. If available, please use the Digital Object Identifier (DOI) reference to acknowledge the use of any of these programs in resulting publications. The DOI reference can be found in the documentation accessible via the links below:
Fine-structure constant measurements|
and long-range wavelength-scale
distortion analysis package.
Data analysis software for systematic|
search of exotic transient signals from
optical magnetometer network.
Highly-scalable hyperparameter optimization software using surrogate modeling and|
Automatic detection of Lyman-limit|
regions in both high and low resolution
Machine learning tools for analysis of|
Distributed Acoustic Sensing data for
Time-frequency and sensor network|
analysis software for urban
Apart from being an avid pianist and hiker, I also enjoy photography a lot.
Below are some of my most favorite pictures taken over the years. I am currently
using a Nikon D7200
camera with a Nikon AF-S DX NIKKOR 18-140mm f/3.5-5.6G ED VR lens. That being said, I really like
how pictures are rendered with the iPhone X although I'm also starting to become
an even bigger fan of the camera performance from the Samsung Galaxy.
One of my main passions in life is traveling and exploring different cultures and societies.
I started traveling fairly late, after getting a scholarship to do my Master's in
Chile. This was a dream come true and the first time I was leaving my family to go
and live at the end of the world, as Chilean like to call their land. I stayed
there for over two years and got to explore South America and its culture. This gave me
a taste for discovering new places around the world and I haven't stopped doing just
that since then. Every stay has been for me an opportunity to experience different
standards of living and have a deeper appreciation for the social, economic and
environmental issues faced by each society. I also truly believe that engaging in
the challenges some communities might have and taking lessons from the solutions
they may come up with could solve decade-long problems in other regions of the world.
Sometimes, there is no need to look for a solution, sometimes the only thing we need
is to look out the window and see what others do... Below is a world map showing all
the flights I took around the world over the last decade (code available
It has been an amazing journey so far!