* Mobile app for finding climbing partners. 
                                * Live on 
Google Play and the 
App Store; 
splash site on custom domain
                                * 
Backend and 
frontend code 
                                * Flutter frontend, backend in Rails + Postgres on AWS Lightsail.
                                * Full chat functionality including read receipts, emoji, hyperlinks, and push notifications
                                * Websockets on Redis + ActionCable, notifications through Firebase Cloud Messaging.
                                * Ads via AdMob, authentication + email confirmation, Google Analytics, Airbrake error tracking, location services, etc, etc.
                            
   
                         
                        
                            
                            
                                * From cargo van to eco-friendly cozy home.
                                * 
Pictures of the build
                                * Full electrical, kitchen, bath, and climate control.
 
                                * 1500w (grid-level, appx) solar & lithium electrical w redundant charging.
 
                                * 120v (standard) house wiring, lighting & wall sockets.
                                * 2 week water capacity & plumbing including sink, shower, & waste tank.
                                * Full insulation & climate control, lived and tested from 100°F to -20°F.
                                * A large, committing project that required endless problem-solving, ingenuity, and creativity- and completed well under a modest budget, at that!
                            
 
                         
                        
                            
                            * Untitled: A natural language processing (NLP) data extraction work project merging data analysis experience with technical skills.
                            * 
The code & README
                            * The goal was to extract thesis topics (indication sentences) from plaintext input (health technology assessment reviews).
                            * Built and trained a novel TensorFlow model and Vector Space Model (Gensim) using the ContextMatter database.
                            * Semantic meaning of words as determined by the model was the only signal available. 
                            * Models were placed in a Flask server in Docker on AWS integrated into an existing web-scraping workflow (Jenkins).
                            * Indications from documents from 10+ countries, languages, and highly variable formats were extracted with 80+ percent accuracy.