Jack's Portfolio

About

I am an engineer at heart with a passion for my work. Growing up in Iowa, surrounded by corn fields and gardening with my family, I developed a love for the environment. My entire career, I've been focusing on wrangling data together for agronomists and environmentalists, scientists, farmers, and business professionals whether on the land or in the sea. Although it's hard to estimate, I've helped in the collection of billions images for computer vision, petabytes of telemetry from sensors all over the world, and trained models with billions of parameters

Here are some highlights from my professional work:

Long-term distributed timelapse solution for phenotyping

Custom stereographic camera with real-time video processing capabilities

Mask-RCNN model trained on hand-gathered dataset highlighting errors of disease on coffee leaves

Data pipeline performing pose detection and stereographic length estimation

Optimized stereographic homography matching for fish tracking

Realtime stereographic depth estimation from an ROV mounted camera

Segmentation, tracking, long-term reassociation, species identification, and counting of fish

Custom CNN+LSTM classification model of high-frequency sounds from marine mammals using spectrograms

GAN based synthetic generation for improved object detection model accuracy

As a life long student, I'm constantly learning and expanding my skillset. The environment will always remain a passion of mine and whether or not I'm working in the field, I will always be working to protect it. If your organization is in need please reach out to me at my email.

Projects

I work on many different projects in my free time. Applying computer vision and AI to coffee plants, fish, scuba diving footage, custom game engines, sales projections for my side hustle, and other querkier projects. Feel free to check out any of my projects you find interesting.

NOTE: When I develop these notebooks, they are exploratory in nature. They may contain errors, incomplete analysis, or failed experiments. I try to showcase my learning process as best as I can, as I believe there is humility and growth in our mistakes. If you find something wrong, please reach out to me so I can correct it!

Other

I've been fiddling with code since 2014. So I've made a lot of different things in that time. My first program was a terminal based vending machine code written in C! And it contained `goto` statements. Yuck! Most of everything I've made goes unpublished in any blog or notebook. But here are some direct links so some of my favorites.

Python library for graphing live data streams in the terminal Rust driver for DHT11 temperature/humidity sensor for Raspberry Pi Rust driver for MCP3008 10bit analog to digital converter for Raspberry Pi Simple 2D Physics simulation with gravity, elastic collision rectification, and seperate axis theorm for collision detection, in Rust with WASM support Ambitious project applying auto color correction and object detection/tracking to beautify scuba diving footage Stereographic camera calibration for tunable depth estimation Rust crate for debouncing functions Rust crate for a window data structure Lazy network protocol for streaming large files out of order and in slow time Rust driver for L293D motor driver for Raspberry Pi 2D Java game engine with quad-tree optimized seperate axis theorem collision detection and elastic recification Unnecessarily verbose modifiable encryption algorithm based on customizable rulebook for personalized usage C++ based light tracking robot simulator with realistic light refraction Very simple TV remote controlled Arduino based robot. (A college project for a science fair) Very first game I ever made that implements network based multiplayer gaming and bullet collision detection and scoring

About

Projects

Feature Extraction and Clustering on Fish in MegaLab Camera

Streaming MegaLab Camera with Community Fish Detector

Spectrogram Generation with Gstreamer and Torch Audio

Manta Ray Identification with Principal Computer Vision Algorithms

Underwater Stereographic Depth Estimation with OpenCV

Ainascan - Tile-Based Semantic Segmentation of Coffee Leaves

Ainascan - Instance Segmentation with Mask R-CNN on Coffee Leaves

AutoFarm - Exploration of Timeseries Models

AutoFarm - Pytorch MaskRCNN Instance Segmentation

AutoFarm - Correct Analysis Of Soil Moisture (Part 2)

Exploring Dominant Color Feature Extractors for ML Applications

Ensemble Methodology for Instance Segmentation Models

Kona Coffee Yield Identification - Research Update

IBM Deep Learning Course - Kona Coffee Yield

IBM Unsupervised Course - Kona Coffee Yield

IBM Classification Course - Kona Coffee Yield

IBM Regression Course - USA COVID19 Anaylsis

IBM Data Science Course - Deep Sea Coral Exploration

AutoFarm - Analysis Of Soil Moisture

Other