Hebbian Robotics
Foundational infrastructure for Physical AI
Hebbian Robotics builds open source foundational infrastructure for Physical AI to develop robots that extend what humanity can do, for the benefit of the communities they serve.
Active Founders
Co-Founder
Brandon is the co-founder of Hebbian Robotics, which builds APIs that help robotics data teams verify data quality for model training. He previously built Gigit AI, a Sequoia-backed AI customer experience startup. His research on AI for low-resource languages topped SEA-HELM, an LLM benchmark co-developed with Percy Liang's research group at Stanford. He briefly pursued a PhD in Robotics at Columbia and NTU before starting Hebbian.
Brandon is the co-founder of Hebbian Robotics, which builds APIs that help robotics data teams verify data quality for model training. He previously built Gigit AI, a Sequoia-backed AI customer experience startup. His research on AI for low-resource languages topped SEA-HELM, an LLM benchmark co-developed with Percy Liang's research group at Stanford. He briefly pursued a PhD in Robotics at Columbia and NTU before starting Hebbian.
Company Launches
Hebbian Robotics - APIs for searching and analyzing Physical AI data at scale
See original launch post****
TLDR;
- Hebbian Robotics builds APIs for searching and analyzing Physical AI data at scale
- Our APIs enable data providers to get on-demand signals and metrics on the quality of their data, without managing infrastructure
- We help our customers evaluate data quality without training a robotics model
Our Ask
- Buyers want data they can trust. If you're selling egocentric or robot data, Let’s Chat (https://calendly.com/brandon-hebbianrobotics/30min) and we'll show you how quality signals and metrics help you sell more at higher prices
- Check out Pareto (https://pareto.hebbianrobotics.com/), an example of how our APIs can be used
Who We Are
- We’re Brandon and Kingston, we've trained AI models for industrial cleaning robots and built production systems for camera fleets across the globe
- Brandon Ong's previous AI startup was backed by Sequoia. He did research on AI for low resource languages that topped Stanford’s LLM benchmark, SEA-HELM, and dropped out of his joint PhD in Robotics at Columbia and NTU
- Kingston Kuan spent years at Jane Street building systems that scaled to millions of messages per second. At Verkada, he worked on their core video surveillance platform that processed large scale multimodal data
The Problem
- High quality data is the bottleneck for unlocking the next frontier of intelligence in robotics
- Buyers want data they can trust
- Data providers don’t have the bandwidth to provide signals and metrics about the quality of their data when they are focused on operational excellence
- This means that model developers are forgoing model improvements due to a lack of high quality data, impeding progress in robotics
Our Solution
- APIs for searching and analyzing Physical AI data at scale that enables providers to get on-demand signals and metrics on data quality
- Buyers demanding per-collector hour limits? Prove compliance while we stay ahead of the workarounds
- Changed standard operating procedures for collection? Monitor for quality drift and regression
- Suspect duplicate samples? Verify quickly with a single API call
- Our APIs enable evaluation of data quality without training a robotics model, fast iteration cycles for high quality data collection, and unlock model improvements
How to reach us
- We want to support data providers who believe in collecting high quality data in an ethical way
- To get started, email brandon.ong@hebbianrobotics.com
YC Photos
