Peter Crocker Email & Phone Number

- Extended Variational Auto Encoders for use in unsupervised clustering of multimodal datasets- Incorporated Distribution-Free, Risk-Controlling Prediction Sets into Cluster Assignment Hardening to provide rigorous confidence bounds around uncertain cluster assignments- Introduced knowledge-driven cluster seeding to guide unsupervised clustering algorithms

- Implemented graph, meta learning, and transfer learning models to approach high-feature, low-sample datasets using Pytorch, Tensorflow, Pandas, Numpy, and Deep Graph Library- Packaged, tested, and deployed models using GitHub, AWS, and Sagemaker under an SKLearn style framework- Analyzed model performance against internal metrics using the Seaborn and.

San Jose, California, US

- Development and Integrated Graph Neural Network based classification and placement algorithms in a Reinforcement Learning framework- Built and Trained GPU accelerated Graphical Neural Networks on Amazon Sagemaker using Deep Graph Library- Achieved state of the art graph classification results on proprietary metrics

Cambridge, MA, US

MIT Course in Machine Learning Applications- Designed and Implemented algorithms to model Bike-Share Network evolution as Natural Language- Developed network "Sentences": encoded sequences reflecting time series evolution of a network- Explored modern machine learning methods through concrete examples and exercises- Focused on model selection and.

Cambridge, MA, US

MIT Capstone on Statistical Analysis- Analyzed large data sets (10M+) using specific, relevant analysis methods and corresponding algorithmic aspects- Engineered a novel approach for behavior-driven surge recognition which became basis for bikeshare surge reduction analysis- Produced a team-based research project: Exploring surge conditions and causes in.

San Francisco, California, US

- Worked alongside the Hardware Verification and Validation developing a Hardware In the Loop test suite for an integral embedded board in Python- Selected out of 100+ interns to present work a public engineering event hosted by Uber- Deployed HIL test suite covering a variety of communication protocols onto Uber ATG's test stack

Cambridge, MA, US

- Contributed to a six person team responsible for bringing an integrated hardware-software system to autonomy- Wrote and implemented algorithms for: Computer Vision, Localization, Path Planning, and Motion Control- Placed 1st with a new school record in the final autonomous race around MIT

Cambridge, Massachusetts, US

- Designed and built a street-legal solar powered vehicle under intense time constraints- Analyzed and Manufactured body components as a member of the aerodynamics and composites subteam- Placed 5th in the 2018 American Solar Challenge, an eight day endurance race across North America

Cambridge, MA, US

- Supported the MIT Knightrider Program: a generalized platform for vehicle autonomy- Tested interface between Autonomy Software and Drive by Wire Commands using C++ ROS nodes

Málaga, Málaga, ES

Worked alongside graduate students at the University of Malaga school for industrial engineering. Our project was to experimentally confirm theoretical calculations on the inverse Magnus effect on a soccer ball with wind tunnel data. Used Labview and force sensors to gather data which we then analyzed with Matlab to determine relative velocities required.

Huntington Beach, CA, US

Seasonal ocean lifeguard. Responsible for the safety of 20-200 patrons a day. Position includes: Ocean rescue, First aid response, CPR training and response, and Public relations.

Master Of Engineering - Meng, Electrical Engineering And Computer Science

Bachelor'S Degree, Electrical Engineering And Computer Science