Dylan Siegel.
Developer & quant explorer
Code. Markets.
Curiosity.
I build software to explore how things work. Lately, that means machine learning and trading systems.
A few things I’m buildingHow to read this
Live order flow drives 512 recurrent units with fast, medium and slow fading memory. Neuron brightness reflects their activity; the three output traces show mean absolute activity in each memory group. Selected connections are shown. The shape is inspired by a fly brain, not a reconstruction of the FlyWire connectome.
The readout measures current best-bid/ask queue sizes and spread. Bid-heavy means over 60% of displayed top-of-book size is on the bid; ask-heavy means under 40%. These are current conditions, not price predictions. The 30-second forecast output has not been trained or validated and remains disabled.
Selected work / 02
Built with conviction.QuantDev.ai
Trading algorithms.
Tested in public.
Sheeplz.com
Find your footing.
Think for yourself.
A little context
Different systems.
Same curiosity.
I spent six years in the U.S. Air Force working on biomedical equipment. Finding out how a system works—and why it doesn’t—was part of the job.
That instinct followed me into software. Now I’m exploring machine learning, markets, and the space where a good question becomes something you can build.
Follow along. Say hello.