Interactive Neural Network Demo
This is a small 2D classification demo powered by a neural network written in C++ and rendered with OpenGL, compiled to WebAssembly and embedded in this Svelte app.
The plot shows how the model separates two classes of points. The summary bar tracks training progress (epoch, loss, accuracy, dataset, optimizer, and weight initialization). Use the panels on the right to step training, enable auto-training, change the dataset, and tweak optimization settings.
Loading neural net...
Epoch 0
Loss 0.0000
Accuracy 0.000
Dataset Two blobs
Optimizer SGD
Init Zero
Auto-train Train 0 epochs
About Me
- I'm a builder who spent two decades writing software, got curious about what the data actually said, and went back to school in my forties to learn the math and research principles. Now I design data systems that don't fall over, from healthcare claims pipelines and LLM evaluations to analytics products. I thrive at companies that need someone to own ambiguous, end-to-end problems.
- Through my consulting practice, Wandering Leaf Studios , I help teams turn messy data into clear strategy. Having worked across Fortune 500s, startups, non-profits, and universities, I'm always chasing the same question: Are we building what actually matters?
- Outside of work, I maintain Open Brewery DB , an open-source dataset and API I built that serves nearly 1 million weekly requests on a pocket-change budget. I'm also a student and graduate TA for Georgia Tech's Online Masters of Science in Analytics (OMSA) program and doing my best to learn in public.