Thesis
Subtle behavior is an underused signal layer.
At BehavLabs, we believe certain subtle human signals are underutilized: the movements of your mouse, your typing patterns, and the smartphone sensors that tell us about your movement.
We are building foundation models for subtle human behaviors. How you handle yourself online can tell us a lot about your current energy level, mood, agitation, attention, and intent.
These signals can be used to improve user wellness, limit doomscrolling, and better understand users in real time. They can also help recommender systems better know when and how to serve content or ads to users.
We believe these capabilities already exist inside well-established players, also known as Big Tech, as closely guarded internal systems. We want to bring real-time behavioral understanding to independent developers, startups, and SMEs.
We aim to train large spatio-temporal, general-purpose behavioral analysis transformers on multivariate time-series sensor and user-input data. We then quantize and distill them to run directly on edge devices, as smartphones become increasingly capable of on-device inference. Learn more about on-device ML.
This would allow independent developers and SMEs to use our proprietary weights and SDKs to power their applications with edge understanding capabilities. Larger and more capable models would be served through an API to analyze specific user traces in greater depth.
Founder
Zach Rodiere
I grew up in a small town in France, studied electrical and computer engineering at ENSEA after going through the classes préparatoires system, and moved to the United States to pursue a Master's in ECE at Georgia Tech.
My background is a mix of applied machine learning, research, and systems engineering. I've worked on transformer models for neural decoding, self-supervised learning for SEEG signals, and GPU/C++ projects ranging from CUDA ray tracing to a market making strategy evaluation engine.
Beyond the technical side, I've consistently gravitated toward leadership and initiative-driven roles. I served as Secretary of the IEEE HKN Beta Mu chapter at Georgia Tech, was Vice President of the business club at ENSEA, and previously ran for student union president.