Wayne Radinsky
1430 S. Jasmine Way · Denver, CO 80224 · (303) 758-1974
https://www.linkedin.com/in/waynerad/ · https://github.com/waynerad

SKILLS

Tools & Technologies:

  • Backend programming languages: Golang (since 2011), PHP, some Java, C++
  • Cloud platforms: Amazon Web Services (AWS), some Docker and Kubernetes
  • Frontend languages/frameworks: JavaScript, jQuery
  • Data science languages/frameworks: Python, numpy, TensorFlow/Keras, R
  • Database technology: relational: SQL language, MySQL, and sqlite3; noSQL: MongoDB

Software architecture design:

Solid engineering background, especially in how to design small software projects so they can become large software projects without turning to spaghetti. I call it “low entropy software.” Every successful large software project ever made started out as a small software project that got larger.

EXPERIENCE

Blue Fire Group, Englewood, CO — Software Engineer (contract)
JULY 2006 — PRESENT

  • Developed three interfaces to three financial transaction systems.
  • Developed multiple data import systems (import real estate data from 4 MLSes and MLS aggregators).
  • Developed a syndication system that exports data to 24 syndication partners.
  • Developed a general-purpose polymorphic ‘Wizard’ system for making wizards in the web UI. Incorporated an import system in a ‘Fast-Start’ Wizard to pull in listings as soon as accounts are created to quickly convince realtors of the value of the system and generate sales.
  • Created a framework for templating (created the framework for professional visual designers).
  • Created a video embedding system, supports videos from 19 video sites and 6 virtual tour sites. Developed production debugging system.
  • Developed video queue system for generation of videos for real estate.
  • Created 29 cron jobs and asynchronous background tasks.
  • Created systems for monitoring cron jobs and background tasks.
  • Managed Amazon AWS resources including EC2 instances, S3 data stores, RDS (databases), and Route 53 (DNS).
  • Back end software is primarily in PHP with some Go, front-end software is primarily JavaScript with jQuery

Anzac International, Livermore, CA — Software Engineer (contract)
OCTOBER 2001 — DECEMBER 2007

Developed PhoneSeal, a Windows application to correct databases affected by telecommunications companies' area code splits. UI wizard was in VB and low-level processing was done in C++.

Apollo Group International, Livermore, CA — Software Engineer (contract)
MAY 2003 — JUNE 2006

Contract developer for TheaterLogics, a home theater project. Developed the setup/configuration system and code for responding to the on screen display (OSD). The setup wizard was done in VB and the OSD and code for communicating with the device driver was C++.

EDUCATION

Deeplearning.ai — Coursera — Deep Learning Specialty — Certificate awarded
AUGUST 2017 — FEBRUARY 2018

  • Course 1: Neural Networks and Deep Learning (logistic regression, gradient descent, numpy)
  • Course 2: Improving Deep Neural Networks (initialization, regularization, bias correction, gradient descent with momentum, Adam optimization, gradient checking, appropriate scale for hyperparameters, implementing batch normalization, softmax, TensorFlow
  • Course 3: Structuring Machine Learning Projects (error analysis, cleaning up incorrectly labeled data, when dev and test sets have a different distribution from the training set, multi task learning, Transfer learning)
  • Course 4: Convolutional Neural Networks (Keras, padding, strided convolutions, car detection, turning fully connected layers into convolutional layers, the YOLO algorithm, intersection over union, non max suppression algorithm, face recognition, neural style transfer
  • Course 5: Sequence Models (recurrent neural networks (RNNs), vanishing gradients, gated recurrent units (GRUs), Long Short Term Memory (LSTM), bidirectional RNNs, word vectors, cosine similarity, hierarchical softmax, sequence-to-sequence models, BLEU (bilingual evaluation understudy) score, BEAM search, attention model, trigger word detection)

Stanford Machine Learning (Coursera) — Statement of Accomplishment awarded
MARCH 2013 — NOVEMBER 2013

Neural networks in Octave (open source MATLAB clone)

Stanford AI Class (the original MOOC) — Statement of Accomplishment awarded
SEPTEMBER 2011 — DECEMBER 2011

A lot of classic (pre-neural network) AI techniques

University of Colorado at Boulder, Computer Science — Attended
SEPTEMBER 1989 — MAY 1993

Classic algorithms and data structures (recursion, greedy algorithms, hash tables, linked lists, search, graph search, etc), discrete math.

LEADERSHIP

Boulder Future Salon — Co-founder, meeting organizer, presenter
NOVEMBER 2006 — PRESENT

Ongoing discussions about future technologies, changes in technology, and general futurology.

PROJECTS

samesync — File synchronization system written in Golang with privacy via AES256 encryption. Demonstrates Go language design & development. https://github.com/waynerad/samesync

wtelnet — Telnet chat server that can host more than 3,000 simultaneous users. Demonstrates skill at massive concurrency in Go. https://github.com/waynerad/wtelnet

UMT — Interactive algorithmic music composition system written in JavaScript. Demonstrates JavaScript design & development skill as well as music theory. https://github.com/waynerad/umt

Backprop and MNIST — Hand-coded the backpropagation algorithm and re-wrote MNIST in Keras. https://github.com/waynerad/backprop and https://github.com/waynerad/mnist

analyzetmdb — Analyzes the TMDb movie database of 10,865 movies (Jupyter notebook, can be viewed as HTML). Demonstrates data analysis and explanation. https://github.com/waynerad/analyzetmdb

osmlyon — Exploration of Open Street Map data from Lyon, France. Demonstrates data wrangling and explanation. https://github.com/waynerad/osmlyon