We make the brain buildable.
Dydema builds AI models that make sense of brain signals, and records the data to train them. Right now, building anything in neurotech means collecting your own data first. That’s where most ideas die. We do that part. Founded in 2025 and backed by Y Combinator.
why we started
Neurotech is short of data.
Language models learned from the whole internet. Brain models have almost nothing to learn from, because recording people is slow and expensive. So we build the models and record the data.
Julian
Julian does the neuroscience. He’s done every step of it himself, from fitting electrodes to training models, and he’s built organizations to do exactly that.
- before
- Co-founded Longhorn Neurotech at UT Austin, and grew it past 200 members
- Founding engineer at Syntra (YC S24)
- research
- Led EEG neurodata teams at Longhorn Neurotech
- fMRI interpretability @ the Dynamic Medical Image Computing lab, UT Austin
- hands on
- The full neurodata lifecycle. Electrode placement through to training models
- Prototype hardware, firmware, and the signal-processing pipeline
Chris
Chris builds the models and everything they run on. Before Dydema he worked on AI where data is scarce, which describes most of neurotech.
- before
- Structures in Artificial Intelligence Lab, University of Houston
- Founding engineer at Syntra (YC S24)
- research
- Deep learning including graph RAG and embodied AI
- Modelling in sparse-data domains
- hands on
- Backend, API and ML/LLM infrastructure
- Production AI systems and security