c o de s lu rry
John Thomas Projects Other Contact
John Thomas, Software Engineer, San Francisco Bay Area

I have 18+ years of experience in software and hardware engineering; I design and develop complex algorithms for image processing and video in both commercial and R&D environments. I am currently the lead engineer for XYZ Color Science creating color science technologies for the TV and display ecosystem. I graduated from UC Berkeley with a B.A. in mathematics (partial differential equations and numerical analysis) and minored in physics (macroscopic quantum effects).

Highlights:
  • Bioinformatics - DNA Comparison tools in Python, data visualization, image processing
  • Improving Video/TV – Bringing improved accuracy, realism and calibration to the TV/Display ecosystem
  • Color science – GPU implementations of novel science
  • Color Based Search Engines – Created search-by-color for fashion and style magazine
  • Nuclear Medicine Imaging Robotics – Seven axis gantry for dual head gamma camera
  • Image Segmentation – Feature extraction from genomic similarity matrices and gene chips
  • Biolistic Particle Accelerator – High voltage sub-micron accelerator for maize biolistics
  • IoT Hardware – MEMS based motion alarm system
Current Toolbox:
  • C++/C
  • OpenCV
  • GPU CUDA/OpenCL
  • MySQL
  • PHP
  • HTML
  • Python
  • Matlab
XYZ Color Science develops a color science technology based on Dr. Ed Granger's work, previously a professor at RIT and senior scientist for Kodak and X-Rite. I've implemented this novel human visual perception based color system in a series of GPU Kernels, C++ applications and MATLAB prototypes to process real-time video for the TV industry. Our process challenges current calibration methods and produces more accurate life-like imagery.

XYZ has also harnessed this technology to create a color based search engine for a leading style magazine. I wrote the full stack implementation using HTML, PHP, MySQL, OpenCL and GPU Kernels. See below for a simplified working version of our color search engine using the open source Yahoo Flickr Creative Commons (YFCC) image dataset.

Other Interests

AI – deep learning, machine learning, speech and image recognition
Building – Houses to furniture, computers to electronic art
Gardening and horticulture – See my greenhouse built from an old deck and re-purposed windows
Painting – Acrylic Painting (2nd year student)
Robotics – Animal mimicry
Bioinformatics – Data mining the biosphere
A Color Based Search Engine

An example of XYZ's color search engine. A MySQL database was created with images pre-processed by GPUs to extract hue, brightness and saturation. JQuery controls select hue family, brightness, saturation, and color percentage. The database has a few hundred thousand image URLs from the Yahoo Flickr Creative Commons (YFCC) dataset.

Try it here!
Computational Genomics/Bioinformatics

Here's a feature extraction algorithm I created to extract similarity features and is an example of a clustering algorithm. Interesting similarity objects are circled in red. The annotations show base pairing, and organism descriptions. This dot plot is between Thalassiosira (a diatom) and Stenocarpella (a plant pathogenic fungus) comparing an evolutionarily conserved S18 ribosomal RNA gene.

Dot Plot image (2000x2000)

Visual schematic of clustering algorithm
Cognative Induced Imagery

Experiments in perception. I've replaced sub-sampled regions with correlated small color images which have similar colors and features.

Two examples, a dog...

and an owl.
Other Projects, Painting and Building

Some of my paintings can be found here and pictures of my greenhouse which was built from an old redwood deck and glass from old double pane windows pulled apart and cut. Peeler core supports, 8 milled 3 light windows, 4 doors. 700-1400 seedlings a year!


Contact Me:

Email:


Subject:


Message: