Ryan Shubert Email & Phone Number

California, United States

California, United States

Pittsburgh, Pa, Us

Software Engineer on the State Estimation team

Pittsburgh, Pennsylvania, Us

Software Engineer on the Tracking team

Cambridge, Ma, Us

Completed my M.Eng. Thesis- "Representation Learning in Visual Navigation for Multi-Agent Drone Control"Redesigned Quadrotor (Drone) Simulator to create a multi-agent reinforcement learning OpenAI gym-like environment. Designing Python (PyTorch) interface to train models in the environment. Augmenting standard algorithms (PPO, SAC, etc.) to support training multiple agents using different recurrent policies in a continuous action space. Training various Neural Net architectures in the environment, namely CNNs, LSTMs, and the novel Neural Circuit Policy. Evaluating each of the architectures based off their performance as well as their task understanding ability. Assessing task understanding through examining the saliency maps each of the models produce with their inputs.

Vienna, Virginia, Us

Continued work from internship. Developed several more data-management functions for our REST API to support CRUD functionality in our PostgreSQL database. Created new methods in the API to support a faceted search on our data. Performed integrations of our API with 3rd party apps like Splunk, Demisto, and Phantom. Used RegEx to parse various different sets of syslogs and extract information. Developed REST service to appropriately route incoming requests to newly added API functions.Agile Development: Participated in daily scrums, demonstrated my work biweekly to team and project manager, and collaborated on Sprint planning

Vienna, Virginia, Us

Developed Protobuf data structures for our gRPC. Integrated our API with our PostgreSQL database. Wrote data-management functions for our REST API to support CRUD functionality in our PostgreSQL database

Under the guidance of Nicola Ferralis of the Grossman Group at MIT, I used Raman Spectroscopy to analyze carbonaceous ancient sediment. By processing the data from the sub-peaks of the spectra, it is possible to calculate the Hydrogen-to-Carbon ratio. This ratio gives insights to the age and preservation of the sample, and can be used to giver a better idea of the origin of that organic material (from decomposition of life or from natural processes). I then used the data I gathered to improve upon a machine learning algorithm for analyzing similar samples. If successful, this algorithm could be implemented in NASA's Mars 2020 Rover in its SHERLOC instrument. During this time, I learned how to operate and process the data from a Raman Spectrometer and got an introduction to machine learning.

Researched the cellular and nuclear entry processes of HIV-1, specifically focusing on the interactions between the Capsid protein and various proteins in the target cell. Used technologies such as site-directed mutagenesis and CRISPR.

Master Of Engineering - Meng, Electrical Engineering & Computer Science (Artificial Intelligence Concentration)

Bachelor'S Degree, Computer Science And Electrical Engineering