Machine/Deep Learning Engineer

Ana Noriega, MS

Atlanta, GA 470-***-**** *********@*****.***

http://www.linkedin.com/in/ana-noriega-33349aba/ https://ananori99.wixsite.com/blog

MACHINE/DEEP LEARNING ENGINEER

Machine & Deep Learning Artificial Intelligence Applied Mathematics Dynamical Systems

Exemplary record in organizational goal attainment through high-quality software system development.

Recent MS in Computer Science graduate with prior MS, BS in Mathematics. Strong theoretical background for understanding, assessing, and implementing data science/ machine learning results. Possess unique background in mathematical and physical approaches relevant to deep learning (including the use of numerical analysis, matrix analysis, complex networks and dynamical systems); capable of leveraging machine learning to build processes for gathering insights from data and driving continuous improvement. Solid record of success in interpersonal communication at all levels, as well as completing research projects efficiently based on Agile methodology. Taught 30 students.

Key Strengths:

Programming Software Engineering Mathematical Modeling Data Analysis Signal Processing Neuroscience Recurrent Neural Networks Weight Matrix Initialization Characterizing Predictability Population Models Natural Language Processing Statistical Physics Project Execution Innovative Persuasive Communicator Problem-solving Deadline Driven

Technical Skills:

Programming: Python C/C++ Scala JAVA LaTeX Matlab SQL

Scientific Computing: Keras Tensorflow PyTorch Scikit-learn Pandas, Jupyter Notebook OpenAIGym Lucene Gensim NetworkX Google Cloud Platform

Computer Science Theory: Data Science Machine Learning Time Series Analysis Simulations Artificial Intelligence Information Retrieval Reinforcement Learning Artificial Intelligence Numerical Analysis Complex Network Sciences (Graphs, Not Hardware)

Other Theory: Numerical Linear Algebra Matrix Analysis Graph Theory Topology Ordinary Differential Equations Mathematical Neuroscience Math Biology Dynamical Systems and Bifurcation Theory Game Theory Economics Topology Cardiac Rythmogenisis, Neuron and Neural Mass Models

Education, Certification & Professional Development

Master of Science in Computer Science (Concentration: Data Science) – Emory University, Atlanta, GA (2020)

Master of Science in Math (Concentration: Scientific Computing) (Missing Thesis) – Georgia State University, Atlanta, GA

Bachelor of Science in Mathematics – Georgia State University, Atlanta, GA (2016) (Recipient of Brains & Behavior Summer Fellowship, Neuroscience Institute)

Selected Highlights

Played a key role in characterizing the workloads of servers at Emory University, focusing specifically on periodicity or predictability; the engineered features ended up being effective; worked collaboratively with three team members.

Performed consistent characterization of weight matrices for optimal performance successfully as part of a semester-long research project on deep learning matrix initialization weights at Emory University.

Professional Experience

Emory University Atlanta, GA 12/17-5/20

Graduate Assistant

Played key role in biologically informed weight initialization of recurrent neural networks while working with Avani Waldani in SymBioSys lab. Served as instructor of record for Summer 2019 course on intro computer science (Java); designed course materials.

Completed significant readings on Echo State Networks and the work of Grossberg in bio-plausible neural networks.

Functioned as teaching assistant for two semesters of introductory computer science; served as substitute lecturer for a 70-student section of the course.

Neuroscience Institute, Georgia State University Atlanta, GA 11/12-5/16

Research Assistant

Used dynamical systems and modeling approaches to learn and verify underlying neuronal circuits that could have produced the electrical recordings of sea slug swim patterns in the Shilnikov lab.

Performed simulation in terms of ordinary differential equations with regard to synchronization and dynamic richness of circuits of oscillatory nodes.

Was part of a large team using experimental (dynamic clamp) recordings of neurons from the swim circuit of the sea slug Melibe to test hypothesized swim circuit topologies.

Completed characterization of chaotic times series’ ”chaoticness” by symbolic dynamics and heuristics from field of information theory.

Projects

Information Retrieval

Assumed responsibility for initial topic modeling by Latent Dirichlet Analysis, corpus prepossessing, and analysis of Natural Language Tool Kit Reuters corpus, followed by graphical analysis of networks for result under multiple ”distance” measures between topics; focused on clustering by Girvan-Newman.

Artificial Intelligence

Examined impact of biologically inspired back-links in reinforcement learning in OpenAIGym in bipedal walker environment.

Time Series Analysis

Compared new tools (symbolic kneading and Lempel-Ziv Complexity) against standard time series analysis tools (including wavelets, Hilbert-Huang Transform, and Wigner-Ville Distribution) for data with both known and unknown events/bifurcations (including seizure onset in a neural mass model) ; verified whether the new tools worked and when they were better (or worse) than standard tools.

Network Sciences

Characterized block matrices based on compartmentalized models of mouse cerebral cortex using graph theory and matrix analysis; other matrix configurations were also examined as controls with regard to properties sought for initialization in deep learning literature.

Machine Learning

Performed work involving the data set of server workloads generated by different sources; engineered a way to characterize predictability of time series to benchmark against a convolutional neural network viewing images of the work load time series.

Presentations

Brains & Behavior, Atlanta, GA (July 2013)

Poster Presenter: “Bursting on the Rebound”

Presented a poster on the network topology of the dynamical system of the three-cell Fitz-Hugh-Nagumo model that gives rise to a behavior of interest to neuroscientists.

Emory SimBioSys, Atlanta, GA (March 2020)

Speaker: “Echo State Networks to Explore Weight Initialization”

Presented a pitch for current research by introducing Echo State Networks as a way in which to focus on initialization of neural network weights.

Contact this candidate