Side projects I've pursued to help me explore & learn
Incredible Times we Live In
I feel blessed to live in a time of such interesting developments -- how does the Chinese proverb go??
Taking some time since my last role (at Crunchyroll), I've recently been able to catch up on & deepen my exposure to recent developments in several of my favorite technologies, including:
- WebAssembly
- Visualization of Real-World (Flight) Data
- Cloud Deployment Options / GCP
- Taking more advantage of what GitHub has to offer
golang,Python,C#,JavaPandas,NextJS,Kafka,KubernetesChatGPT,LangChain,Chroma
See more about some of the projects I've pursued in the past in which I leveraged these:
Kafkloud
I built a general-purpose data pipeline named kafkloud in
a monorepo using
multiple languages/frameworks, supporting multiple deployment targets including
local, Docker Swarm and
Google Kubernetes Engine (GKE).
gochat
Ever since ChatGPT
came out, I've fiddled with being able to "plug in" features to (like
a lot of other people!) - and help myself learn more about - this
fascinating interactive paradigm.
Instead of writing my own "plugin" for ChatGPT, I've
tried to stay platform-neutral (sense that I avoid commitment?), so
I created gochat,
an open-source CLI-based chat loop (written in golang)
which sits on top of several popular LLM platforms, and supports
additional configurable features - with more planned!
Flight Visualizer 
Out of my love for both aviation and leveraging the value of data, I wrote and released
an application in golang called flightvisualizer which makes it
easy to create a "virtual replay"
of any flight taken in the real world (within the past week or so) by querying
FlightAware's
AeroAPI.
Just give it the "tail number" (identifier) of the aircraft and the date/time of the flight, and it will replay it in Google Earth, along with (optional) clickable performance measurements reported along the flight path.
MSFS2020-PilotPathRecorder
Pilot Path Recorder is similar to Flight Visualizer, but is focused on replaying and enhancing the experience of simulated (rather than real-world) flights -- either in real-time, or de-brief.
It's written in C#
InnovAIteLab
I couldn't resist the temptation to "get my hands dirty" with several of the AI open-source APIs in the news these days, such as from OpenAI, LangChain and Chroma.
Being able to use natural language to ask questions to find, recall and make sense of years of notes and saved documents is truly valuable.
Witnessing the ease and value of using this technology to set up and turn a Python "ingester" loose on large, unorganized collections of files (i.e., create embeddings) thereby enabling this use case is pretty impressive.
