Experience

Google Colab

I've been using Google Colab since before it was released publicly in 2017. It's been a great tool for sharing and pairing on machine learning projects like computer vision for Cainthus, GreenThumb IO, and nSightSurgical.

iTerm2

Sometimes iTerm2 is great for special usecases like typing, or hitting return, on multiple tabs at once.

Jupyter

I used to use plain old Jupyter notebooks, but mostly use Google Colab these days. Sometimes I will run notebooks on remote containers with the VSCode extension.

Notion

Documentation is critical for async processes and notion is a great way to collaboratively produce documentation.

Slack

Slack is a great way to partially sync up with the expectation being you don't have to respond imediately. Chat is self documenting in a way because public channels record ideas and decisions from disucussions that are searchable for posterity. You can easily use integrations to take a message and create a task/doc or just copy paste.

MailGun

I've used MailGun and SendGrid to send emails to users in dozens of products

RabbitMQ

At Cainthus we tested using RabbitMQ with Celery, but ultimitely Redis is my preferred message broker for it's simplicity and powerful feature set.

ZeroMQ

I've used ZMQ for IPC (Inter Process Communication) to share image data between processes on the same machine. I've found using shared memory objects with Faktory to be more performant and more concise to develop with. Faktory also provides the extensibility to use different languages for jobs, like API clients in JavaScript as well as processs jobs on distributed machines.

Tensor Flow

I've deployed ML pipelines that train and use nueral networks for computer vision inference on image data.

TypeScript

I have done audits for and deployed a few applications in TypeScript, but it's not my daily driver. Typing great, but it's not an excuse to skip testing.

PHP

I have done audits for and deployed a few applications in PHP, but it's not my daily driver.

Rust

Rust is pretty sweet and I look forward to doing more work with it.

Elixir

Elixir and the Pheonix framework have had my interest for a while, but I've limited production experience.

Stimulus

Stimulus controllers are just enough JavaScript for me.

Tailwind CSS

Tailwind is a great CSS utility framework that makes me feel like I'm not terrible at CSS. This site uses the Tailwind UI Spotlight template as a base with minor customizations.

Alpine.js

Alpine is great, lightweight and there's even a sprinkle of Alpine in this site!

PostCSS

PostCSS is great for processing CSS with plugins like Tailwind CSS and Autoprefixer.

Sinatra

I've built a few microservices with Sinatra to expose APIs with Ruby. One of the most interesting examples that made it to production was a geocoding service that took coordinates and returned location names using OSM, PostGIS, and Redis for a caching layer. It could handle batches of coordinates with a custom stored proceedure and serviced tens of millions of requests a day.

Flask

At Cainthus we served our APIs with Flask for our React apps that rendered data derived from Computer Vision

Express.js

At Momentage we created a simple file upload service with Express JS and MongoDB's GridFS for resumable file uploads with S3 and CloudFront serving the uploaded assets.

Django

At Cainthus we used Django for our Drone mapping web app with Celery for background jobs to process map tile analysis and visualizations.

CakePHP

At Focus Testing Services we had a CakePHP JSON API serving our desktop app that was built with Flash FlexBuilder using ActionScript 3

Laravel

Laravel is cool in my book.

GraphQL

I've used RESTful APIs for over a decade and starting working with GraphQL in 2017

Apollo GraphQL

Microsoft SQL Server

At CATS we used Microsoft SQL Server. I help to maintain the Windows Server machine's hardware (SCSI RAID, UPS battery backups, etc.) and software upgrades (Windows System and Applicaiton Dependencies). I also had an opporunity to contribute to the development of our Java JBoss web app and services for the backend with a good old JDBC Microsoft SQL Server driver and Swing Desktop frontend. Coincidentally it was my first time using SQL and last time using Microsoft SQL Server.

OpenStreetMap

OpenStreetMap is a fantasic open source data and tool set that provides multipolygon information about administrative regions, streets, and buildings around the world. It was the foundational dataset used in the Banjo reverse geocoder service that I built in 2013.

Heroku

I started using Heroku is 2012 and I still really enjoy Heroku to this day. I should probably checkout some of the new hotness like Render, Vercel, Fly.io, and the like, but I mostly just use AWS ECS/EKS with Terraform and GitHub Actions

Linode

I've managed a couple of client's deployments on Linode, but it's definitely not my preference.

Apple Silicon

I've been able to leverage the Apple Silicon architecture using CoreMLTools in python to leverage the Apple Neural Engine for on-device inference. I've also used OpenCV and NumPy with Silicon support for GPU computations.

Raspberry Pi

I have been tinkering with Raspberry Pis for over a decade now. Anywhere from heliostats to monitoring plants with computer vision to water them

Ubuntu

I've been using Ubuntu for well over a decade now. I've used Ubuntu Desktop for development as well as Ubuntu Server for cloud and on-premise deployments