Post on 13-Jan-2017
Rory Smith, Ph.D.
California Institute of Technology • 1200 E. California Blvd • Pasadena, CA• 91125 • MC 100-‐36 • 747-‐238-‐8603 e-‐mail roryjesmith@gmail.com • linkedin https://www.linkedin.com/in/roryjsmith
SKILLS Signal Processing Data Analysis Bayesian Inference
Data Compression Stochastic Sampling Mathematical Modeling Approximation Theory Parallel Computing Multi-‐parameter Interpolation Python C/C++ Scalable Algorithms LaTeX (non)linear regression Scientific/Technical Writing
EDUCATION Ph.D. Physics (2013) University of Birmingham, U.K. MSci. (Hons) Theoretical Physics, Class 1 (2009) University of Birmingham, U.K.
EXPERIENCE
California Institute of Technology Senior postdoctoral researcher in physics (Current) and postdoctoral researcher in physics (2013-‐2016)
! I am a member of the data science group in the LIGO laboratory at Caltech. My research is at the interface of mathematical modeling, the development of low-‐latency and scalable Bayesian inference algorithms, signal processing and machine learning. For example, I am: • Leading an effort within the international collaboration LIGO to design and implement
scalable, low-‐latency matched filtering and Monte-‐Carlo inference algorithms to extract astronomical information from noisy time-‐series data. My work has improved the efficiency of Monte Carlo sampling algorithms for inference by several orders of magnitude, increasing the feasibility of making astronomical measurements using LIGO data. [http://journals.aps.org/prd/abstract/10.1103/PhysRevD.94.044031]
• Designing and implementing algorithms to improve the quality of LIGO data – and
astronomical measurements using LIGO data – through linear and non-‐linear regression of noise sources.
• Producing tools to simulate realistic interferometers in near real-‐time for experimental design and commissioning. [http://iopscience.iop.org/article/10.1088/2040-‐8978/18/2/025604]
! The links between these areas are a set of techniques known collectively as “reduced order modeling”. These techniques include, but are not limited to, finding highly compressed expressions for likelihood functions that arise in Bayesian inference, which utilize multi-‐parameter interpolation with sparse and scattered data.
! I am a contributor to two software libraries used by the LIGO Scientific Collaboration: The
inference libraries in LIGO’s data analysis software suite (LALsuite), and the open source interferometer simulation software Finesse. These libraries are routinely used by a collaboration of several hundred people.
FASTech LLC Co-‐founder and consultant in data science (2014-‐2015)
• I co-‐founded FASTech (Fast Analysis and Simulation Technologies) LLC, a startup based in San
Diego, to explore commercial applications of reduced order modeling and signal processing/inference techniques that I developed as a postdoc at Caltech. Such applications include data compression, efficient inference on large data sets, and simulations of stochastic physical processes. As a consultant, I help draft a SBIR (Small Business Innovation Research) grant proposals for to develop simulations of jet flows using reduced order modeling. In addition, we also explored and formed strategic partnerships with DAC (Decisive Analytics Corporation) and FiskeTech, two companies specializing in predictive analytics for defense.
University of Birmingham Ph.D. in physics (2009-‐2013)
! My thesis focused on efficient methods to detect astronomical signals and perform inference on
those signals in data from the LIGO experiment. My thesis received an honorable mention in a PhD thesis award from an international committee of physicists and astronomers. [https://gwic.ligo.org/thesisprize/2013/]
! I was also the recipient of a (competitive) Canadian visiting graduate fellowship award at Perimeter Institute for Theoretical Physics (2012) where I spent six months working in the gravitational physics group on data analysis techniques for the LIGO experiment.
Publications and Presentations
An up-‐to-‐date list of my publications can be found on Google Scholar: https://scholar.google.co.uk/citations?user=db687GsAAAAJ&hl=en&oi=ao
Caltech • 1200 E. California Blvd • Pasadena, CA • 91125 • MC 100-‐36 • e-‐mail roryjesmith@gmail.com