Pete Florence
Co-Founder CEO at Generalist
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPete Florence is a robotics and computer vision researcher-turned-founder with a decade of experience building ML-driven manipulation and 3D perception systems; he holds a PhD from MIT's CSAIL Robot Locomotion Group and currently leads Generalist in San Francisco. He has served as a senior/staff research scientist at DeepMind working on large multimodal models, self-supervised 3D learning, and policy learning for robotic manipulation. An active open-source contributor, Pete has made practical engineering contributions to high-profile projects such as the Drake robotics library, the MIT underactuated course text, and PyTorch Dense Correspondence (LabelFusionDataset and domain randomization), bridging research code and production tooling. His career arc is unusually interdisciplinary—starting in chemistry and physics (Princeton, Cambridge) before pivoting to EECS—giving him a broad scientific lens that informs both product strategy and hands-on systems work.
11 years of coding experience
10 years of employment as a software developer
Ph.D., Electrical Engineering and Computer Science (EECS), Ph.D., Electrical Engineering and Computer Science (EECS) at Massachusetts Institute of TechnologySaratoga High SchoolMaster of Philosophy (MPhil), Physics, Master of Philosophy (MPhil), Physics at Cambridge UniversityA.B., Chemistry, A.B., Chemistry at Princeton University
Stackoverflow
4,812reputation
252kreached
1answer
1question
html
top-1%
webkit
top-1%
ios
top-1%
Github Skills (19)
pytorch10
robotics10
ios10
documentations10
python10
machine-learning10
matlab10
webkit10
data-preprocessing10
drake10
html10
computer-vision10
documentation10
data-augmentation9
urdf8
Programming languages (12)
Github contributions (5)
RobotLocomotion/pytorch-dense-correspondence
Feb 2018 - Nov 2019
Code for "Dense Object Nets: Learning Dense Visual Object Descriptors By and For Robotic Manipulation"
Role in this project:
ML Engineer
ML EngineerContributions:606 commits, 20 PRs, 48 pushes in 1 year 9 months
Contributions summary:Pete primarily focused on developing and refining the `LabelFusionDataset` class, a crucial component for loading and preprocessing data from the LabelFusion dataset. Their contributions included implementing and fixing functionality within the dataset class, such as addressing issues with data augmentation and incorporating both the original image and the corresponding mask for training and evaluation. The user also implemented a domain randomization function and worked on integrating tools for plotting and visualizing correspondences, demonstrating an effort to enhance the data processing and analysis pipeline within the context of the PyTorch-based deep learning project.
manipulationroboticsvisionnetsrobotic-manipulation
RussTedrake/underactuated
Aug 2015 - Feb 2018
The course text for MIT 6.832 (and 6.832x on edX)
Role in this project:Technical Writer
Technical WriterContributions:17 commits, 6 PRs, 3 pushes in 2 years 7 months
Contributions summary:Pete primarily focused on improving the documentation of the `underactuated` repository, which serves as a course text. Their contributions included fixing typos and formatting issues within the HTML documentation. The user also updated the documentation to reflect the use of quotes within the HTML and adjusting the example file paths. These edits enhance the clarity and accuracy of the course materials.
edxpythonmit
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.