HI, I'M JOEL
Electrical turned AI/ML Engineer
AI and Machine Learning engineer with a knack for problem solving, debugging, and inventive ideas.
Below is a showcase of AI/ML projects, web apps, and hardware projects I've built. I hope you enjoy!
AI/ML Projects
Some fun projects I did on my own to continue to hone my AI skills in Python, TensorFlow, and Keras.
AI / ML / SOFTWARE SKILLZ
Some of the technologies I'm comfortable with..
After 13 years designing computer chips, writing firmware, and designing electronics, I switched careers to AI and Machine Learning.
My superpower is problem solving, which I achieve through my ability to learn quickly and research thoroughly.
Since AI and ML are also superpowers for problem solving, they make a symbiotic and powerful extension of my abilities.
Some Web Apps I've Made
Spaywall.com
A website and browser extension to help make online news more freely accessible for all, and legally!
Built to Scale
Runs on JavaScript in the cloud via Cloudflare Workers. Browser extension is client-side JavaScript.
Loved by Users
The most popular paywall extension in the Chrome store. Over 35K active users and growing.
Always Improving
You know you've made something good when the copycats come crawling..
Archived Web Apps
Various projects or bootstrapped startup ideas that didn't make it, for one reason or another.
Haven Bugouts
Custom storefront related to the survival gear niche. The site allowed you to create bespoke survival kits in a custom-built shopping cart and then passed you to checkout on Amazon. Two person project. Python/Django, Heroku, background workers, Amazon Product Advertising API, HTML/CSS/JavaScript.
Checkmate
Developed a web app that tracks the progress of circuit board designs through various checklist milestones. Database-backed CRUD app allowed users to add and remove blocks, assign block owners, and mark progress on each checklist item across multiple blocks. Python/Django, AJAX, HTML/CSS/JavaScript.
KnewsStand
One digital news subscription that would get you access to all partner sites. I developed a proof of concept that let you log into the partner sites (similar to Single Sign-On). The difficulty of partnering with giant conglomerates and differences in co-founder expectations pretty much doomed this one from the start.
Hardware
My first love
I made an EV
C-RexThe conversion of my Honda CRX to battery-powered EV. Over 300 hours put into engineering, custom metal fabrication, circuit design, and high-voltage electrical work.
C-Rex was faster than stock, cost less to operate, had fewer moving parts, and was more fun to drive! Sadly, I sold him when I became a dad since he had no backseats. ðŸ˜
Gutting the internal combustion parts
The easy part.. if it leaks or squeaks, it's gone!
Installing the electric motor
DC motor mounted to the original transmission
Batteries arrived (to my great relief)
Still have that fresh-from-Shenzen smell!
Strapping the battery packs
A good chunk of time went into designing the battery packs, the heart of any EV
Rear battery pack
I cut a hole in the floor behind the front seats. GR8 Stuff is my friend.
Installing front battery racks
Painted maize, of course
Electric drivetrain finished
Six months of planning, setbacks, and late nights
Full body, wheel, & glass restoration
Lots of body work and a brand new paint job later
I made a wrist-worn motion-activated stungun
StunlyAfter my sister had a scary encounter in the park, I set out to make a stungun that could be operated hands-free, so there was no excuse to leave it at home.
An onboard accelerometer detects an intentional double-shake of the device to set it off. I designed the circuit and PCB, drew the enclosure in CAD, and wrote the firmware.
The final product worked, but my wife was too scared to have such high-voltage so near her wrist. And that's the tale of how I learned the hard lesson of not talking to my users first.
Teeny tiny parts
The biggest challenge of this project was making the damn thing small enough
Baking an early prototype
I didn't have a reflow oven, so I made one with a toaster oven and a kit
Measuring and plotting accelerometer response
Plotting the accelerometer readings in Python to dial in the best settings
The profile of a double-shake
We want it to trigger during an intentional shake, but not from normal running
Running in the lab like an idiot
Strapping a wired-in early prototype to my arm and jogging in place to get live readings
Soooo maaany batteries..
Tested batteries on the bench across dozens of manufacturers (pro tip: manufacturers lie)
An enclosure rendering
Things always look so much bigger on your screen than when 3D printed
You want renderings? I've got plenty
But who cares, no big deal, I want morrrre!
I made an internet-connected car alarm
BMMPRAs a consultant, I turned my client's idea for an internet-connected impact-sensing car alarm into a shipping product. He's sold millions in revenue.. wish I'd asked for equity 😂
BMMPR plugs into the OBD-II port, and has cellular and bluetooth connectivity, GPS, and an accelerometer. I designed all the hardware and wrote the firmware to connect to the mobile app and an AWS backend.
Breadboarding the prototype
The journey of 1000 miles begins with one giant breadboard covered in evaluation boards
Debugging GPS lock
Using a development board to figure out why I wasn't getting GPS lock
BMMPR mainboard
The mainboard contains the OBD connector, buck converter, accelerometer, and cellular radio
BMMPR daughterboard
The daughterboard plugs on top of the mainboard and has the microcontroller, bluetooth radio, and GPS module
Online Courses
Learning whatever skills I need to solve problems.
TensorFlow Developer Certificate
Google-certified TensorFlow Developer Certificate. Foundational, practical machine learning skills via the building and training of models using TensorFlow.
TensorFlow Keras Regression Classification Image Classification NLP Time Series
DeepLearning.AI's Natural Language Processing Specialization
Foundations of NLP covering logistic regression, naïve Bayes, hidden Markov models, word embeddings, recurrent neural networks, Siamese networks, encoder-decoder models, attention, and transformers.
Autocomplete Sentiment Analysis Shakespeare Generative RNN Question Duplicates Seq2Seq Chatbots Summarizers Machine Translation
DeepLearning.AI's Deep Learning Specialization
Fundamentals of Deep Learning taught by Andrew Ng. Identify key architectural tradeoffs, analyze datasets, build deep neural networks in TensorFlow, build CNNs for image recognition, build and train RNNs and HuggingFace transformer models for NLP tasks.
NN from Scratch Face Recognition YOLO Image Segmentation Sentiment Analysis NMT Trigger Word Detection TensorFlow
Stanford's Machine Learning Specialization
Fundamentals of machine learning taught by Andrew Ng. Supervised machine learning algorithms, including linear regression, logistic regression, neural networks, decision trees, and reinforcement learning in Python.
Handwritten digit recognition Spam classification filter Anomaly detection Movie recommendation engine
Harvard's CS50 Web Programming with Python and JavaScript
Design of web apps with Python, JavaScript, Django, SQL, and Bootstrap. Topics include database design, scalability, security, and user experience.
Review site Slack app Shopping cart
Udacity's Full-Stack Web Development Nanodegree
Learn to build database-backed APIs and web applications with Python, Flask, and JavaScript. Database design, manage user authentication and access control with Docker and Kubernetes.
Music venue app API design Roles-based access control
They said it, not me..
“It is without hesitation that I give Joel my highest praise”
“Joel performed well beyond my high expectations. He was self motivated, and was quickly able to set his own goals and timelines for his work. He did excellent independent research ... He incorporated both large and small pieces of innovations to meet the demanding project requirements. Joel maintained a high level of focus in his day to day work. He is not easily distracted from the task at hand. I cannot remember him ever missing a deadline for a deliverable. Joel worked very well with the other engineers in the group. He works well under pressure and is able to bring his ideas in team discussions in a positive, non-confrontational manner. He delivers across all possible considerations: productivity, initiative, responsiveness , and professionalism. His friendly positive personality is a great asset to any team environment.”
Thomas Grutkowski
Principal Engineer at Intel
“We still talk about some of those early days where you really saved our bacon”
Thomas Reilly
CEO at Access Sensor Technologies
“Managers and technical leads knew they could trust him”
“Joel did well working independently, even on difficult problems ... His work was of the highest quality. Throughout his time on my team, Joel showed a knack for learning new subjects quickly, he showed excellent communication/teamwork, and he reliably met his schedules. I'd be happy to have Joel return to my team!”
Paul Thayer
Director of Engineering at Intel
“Passionate, precise and creative”
“Joel understands complex matters even when outside of his direct area of expertise. He is quickly able to learn new software tools, which enables him to spend less time with the mechanics of the job and more time doing the actual work before him. Joel's work ethic and attitude are an asset to any team.”
Mike Barber
Principal Engineer at Edge Scientific