Experience

• 

  Lightmatter

  • United States
  • Computer Hardware Manufacturing
  • 100 - 200 Employee

  • Machine Learning Scientist

    • May 2021 - Present

  • Principal Solutions Architect

    • May 2022 - Present

    I meet with customers to understand their computational workloads and collect their requirements. Simultaneously, I introduce them to technical details of Lightmatter's photonic chip technology and demonstrate how it can be used to accelerate both their machine learning and signal processing applications. I am responsible for the design, description and managing the development of solutions to support our clients' business strategies. Often, this requires that I use my ML and signal processing background to code solutions or prototype solutions for software engineers. I work with our sales team to design contracts and collaboration goals that are mutually beneficial.

• 

  Lawrence Livermore National Laboratory

  • Livermore

  • Machine Learning Researcher

    • Oct 2014 - Apr 2021
    • Livermore

    I create machine learning and signal processing algorithms for imagery (RGB,LiDAR,SAR), sensor systems (IoT), natural language processing and medical applications. Currently developing deep reinforcement learning models on cancer and sepsis simulations to optimize treatment policies. Also applying GANs to physics simulations to learn better reduced order and surrogate models for design optimization. I code and train convolutional, recurrent and attention-based neural networks. The software tools that I use for this are TensorFlow, PyTorch and Docker for containerization and deployment. I manage projects by setting the technical direction and delegating team member responsibilities. I work closely with end users to determine project scope, data pipelines and technical requirements.

• 

  SRI International

  • Menlo Park, CA

  • Senior Researcher

    • Aug 2009 - Oct 2014
    • Menlo Park, CA

    I create mathematical models and develop algorithms to perform detection and classification for sensor data. Sensors and data include harmonic radar, synthetic aperture radar, LiDAR, RGB images, gas chromatography-mass spectrometry, acoustic and seismic data. Many of my detection algorithms use statistical signal processing methods such as maximum likelihood estimation, hypothesis testing using t-tests for high dimensional data and regression. Classification methods include machine learning algorithms such as Bayesian networks, kernel eigenmap methods, support vector machines, neural networks and decision trees. Experience numerically modeling PDE (mostly related to seismology). Also use compressive sampling for sparse representations and solvers that use interior point methods for optimization such as linear programming and convex optimization.

• 

  U.C. Davis

  • Davis, CA

  • Krener Assistant Professor

    • Jun 2008 - Jul 2009
    • Davis, CA

    I used diffusion magnetic resonance imaging data to construct 3-dimensional models, tractography, for white matter in the human brain. The white matter corresponds to axons. Each axon connects its neuron to other neurons. The goal is to be able to model the network of axons for medical purposes. Investigating ideas that allow local diffusion measurements to build a global model.

• 

  J. Craig Venter Institute

  • Rockville, MD

  • Bioinformatics Analyst

    • May 2008 - Aug 2008
    • Rockville, MD

    Project involved modeling quorum sensing in Burkholderia pseudomallei, Burkholderia mallei and Burkholderia thailandensis. Built an SQL database to store microarray results, wrote perl scripts to automatically load microarray data into the database. Performed t-tests and other statistical analysis to determine which genes were significant in regulating quorum sensing. Created a gene expresson matrix from the microarray results and performed various classification and clustering algorithms such as principal components analysis, nearest neighbors and tree graphs to understand relationships amongst the genes.

• 

  The Norbert Wiener Center

  • Univ. of Maryland at College Park

  • Research Assistant

    • Jan 2005 - Jan 2008
    • Univ. of Maryland at College Park

    I worked with the National Geospatial-intelligence Agency on dimensionality reduction for hyperspectral data. Special emphasis given to kernel eigenmap methods such as Locally Linear Embedding, Laplacian Eigenmaps, and diffusion wavelets. Implemented Nystroem methods for large scale problems. Developed new approaches to kernel construction and endmember selection using finite, unit- norm, tight frames that give sparse representations. Coded algorithms in MATLAB and C++ for dimensionality reduction.

• 

  Napa Valley College

  • Napa, California

  • Mathematics Instructor

    • Jun 2001 - Dec 2005
    • Napa, California

    I taught calculus I-II, differential equations, multivariable calculus and linear algebra.