Danny J Voils

acxmch@r.postjobfree.com www.voils.us

Profile:

Computer engineer who went back to school to acquire new skills. A life-long learner, nerd, doer and innovative thinker. Passionate about unraveling difficult and complex problems. Background in Hardware, Neuromorphic Computing, Software, Cloud Computing, Machine Learning, IoT, Image Processing, and Test.

Technical Knowledge

Machine Learning: Decision trees, Naive Bayes, k-Means, Vector Quantization, Principle Component Analysis, Genetic Algorithms, Reinforcement Learning, Bayesian Networks, Support Vector Machines

Neural Networks: Convolutional Networks, LSTM, Multi-layer Perceptrons, Gradient Descent

Image Processing: Image pyramids, Convolutional filtering, Sparse Coding, Unsupervised Feature Learning

Computer Skills

Languages: C/C++, Python, Perl, Matlab, Verilog, VHDL, x86 assembler

Platforms: MPI, Cuda, OpenCV, Theano, Scikit-Learn, Orange

Web: HTML, CSS, LAMP, AJAX, jQuery, JavaScript, PHP, MySQL, Json, XML

Frameworks: Flask, SQLAlchemy

Software: Ansible, Jenkins, Jira, MS Office, GIMP

Tools: GDB, GIT, CVS, SVN, GRUB, unix shell, regex

OS: Ubuntu, Fedora, Windows

AWS: EC2, S3, Cognito, Lambda

VM: VMWare, QEMU

Employment

Consultant TugboatAI, Portland, OR 2016-current

Small Automated Global Marketing startup. Developing secure cloud framework, test/production environments and back-end architecture. Deployed a secure cloud based test environment using Digital Ocean with significant cost savings compared to AWS. Employed firewall using using Netfilter/iptables. Set up Flask server using uWSGI. Set up cloud infrastructure using Ansible. Using the BLOCKS framework, developing deep recurrent neural network using Nvidia GPUs on AWS to serve back end for a cloud-based chat-bot. Set up AWS user pools with Cognito. Deployed AWS Lambda functions using Gordon.

Developer Exxeno LLC, Portland, OR 2015-2016

Small Internet of Things startup. Responsible for AWS deployment, Software test, Data Analysis and visualization. My work graphing newly acquired data showed proof-of-concept that was key to receiving early funding. Performed QA testing of cloud based services using Python and nosetest framework. Set up test environment using VMware and Ansible. Set up automated testing through Jenkens. Debugged errors and drove fixes to resolution.

Research Assistant Portland State University, 2009-2015

Responsible for research and development of modular associative memory architectures. Responsible for developing software technology and applications based on principles of the neocortex.

In addition to coursework in Machine Learning, Artificial Intelligence, Neural Networks, and Evolutionary Computing, weekly participant in Adaptive Systems Seminars with Melanie Mitchell, Bayesian Methods with Dan Hammerstrom, and Circuits and Algorithms with Merek Perkowski.

Developed a novel modular associative memory model that used ideas from Information Theory to determine the degree of inter-module connectivity. Developed a C++ implementation of Nuementa's Heirarchical Temporal Memory model (HTM). Pioneered the use of coupling HTM with Poggio's HMAX model for image processing and object recognition.

Intel Non-Boolean Systems 2011-2015

Designed a novel modular associative memory model used for image retrieval and storage using Matlab. Used a sparse-coding algorithm to store image features in an associative memory. Used k-Means clustering of HMAX image features to create a sparse-coding dictionary. Stored image features using Hopfield and Willshaw-Palm associative memory array models. Developed a novel image classification scheme based on stored exemplars. Studied, used and wrote about an oscillatory neural computer developed by Intel researcher Dimitri Nikonov.

Invited researcher 2012, 2013

Pazmany Peter Catholic University Faculty of Information Technology, Budapest, Hungary. Worked with other graduate students to develop an image processing and object recognition system using Spin Torque Oscillator Cellular Array models. Used HMAX to extract image features. Used a Genetic Algorithm to find optimal oscillator spacing for efficient storage and recall of image data.

DARPA SYNAPSE 2009-2012

Worked with Huges Researech Labs. Designed a clock accurate C++ neural network routing fabric logic simulator. Pioneered the use of packets to distribute spike trains. Developed an application employing a Python front-end coupled with a logic simulator back-end communicating through a computer network using a multi-threaded cuing mechanism. Supervised engineers to implement routing fabric on a set of 4 FPGAs. Designed an FPGA based neural network simulator created in Verilog employing Izhikevich's spiking neural model.

Design Engineer Mathstar, Portland, OR 2007-2008

Worked in pre-silicon validation of Mathstar's Programmable logic chip. Wrote test plans, designed tests during the early design stages of chip development. Worked on design entry tools. Developed a Verilog design entry methodology for MathStar's programmable logic chip.

Component Design Engineer/Product Engineer, Intel Corporation, 1998-2007

Worked on pre-silicon validation of microprocessors. Wrote test plans, designed tests and monitored logic coverage from early design to tape-out. Also worked on manufacturing test. Incorporated open source software into CPU manufacturing saving the company $50,000 + in licensing fees. Adopted across the company, it resulted in even greater savings. Wrote assembly language programs that found logic bugs in the Pentium III microprocessor during early manufacturing stages.

Education

PhD Electrical/Computer Engineering, all coursework/research complete. 2015

Portland State University

Adviser: Dr. Marek Perkowski

Dissertation: "Functional Decomposition of Associative Memory Networks"

M.S. Electrical/Computer Engineering, 2012

Portland State University

Adviser: Dr. Dan Hammerstrom

Thesis: "Scale Invariant Object Recognition Using Cortical Computational Models and a Robotic Platform"

B.S. Electrical/Computer Engineering, 1995

Iowa State University

Publications and Presentations

•Nikonov, Dmitri E., Gyorgy Csaba, Wolfgang Porod, Tadashi Shibata, Danny Voils, Dan Hammerstrom, Ian A. Young, and George I. Bourianoff. "Coupled-Oscillator Associative Memory Array Operation for Pattern Recognition." Exploratory Solid-State Computational Devices and Circuits, IEEE Journal on 1 (2015): 85-93.

•Andras Horvath, Tamas Roska, Attila Stubendek, Danny Voils, Fernando Corinto, Gyorgy Csaba, Wolfgang Porod, Tadashi Shibata, Dan Hammerstrom, and George Bourianoff. O-cnn vice spin torque oscillator cells and cellular spin-wave interactions in an associative memory. IEEE Transactions on Nanotechnology, 2012.

•Poster: "Modular Associative Memory Networks", ECE PhD Day, Portland State University. Voted best poster. May 2015.

•"Associative Memory Networks for Beyond Boolean Computing" (2014), Intelligent and Adaptive Systems Seminar, Portland State University.

•“Spin Torque Oscillator Boot Camp" (2014), Intelligent and Adaptive Systems Seminar, Portland State University.

•"Introduction to Neural Networks and Unsupervised Learning", (2014) Intel Corporation Non-Boolean Computing Initiative.

Other:

IEEE Member

7 Year Veteran US Navy, E6. Honorable Discharge.

Contact this candidate