Jeremy Cohen Email & Phone Number

Basking Ridge, New Jersey

- Technical lead on Peraton Labs OCCAM program, a TA2 performer for the DARPA ConSec program. Implemented agile team methodologies to manage a large Python, Java, and JavaScript code base with components from CS domains such as Machine Learning, Satisfiability Modulo Theories, sBlack Box Optimization, and UI. Interfacing with other ConSec performers on a weekly basis to maintain technical cohesion and alignment with DARPA goals. Regularly designing and leading implementation of improvements to the scaling of the optimization engine.- Technical and scientific contributor at both the design and implementation levels of a large-scale, modular, graph data modeling system for another DARPA project.- Co-author in a successful bid on the IARPA project TrojAI; helping to develop novel techniques for detecting back doors in pre-trained neural networks without access to large amounts of training or validation data.- Co-author in a successful bid on the DARPA TMVD program. Aiding in the understanding of universal adversarial perturbations.- Co-author of a bid for the DARPA GARD project. This effort involved the primary authorship of a paper showing strong theoretical results in the field of certifiably adversarial resistant neural networks.- Performing ongoing theoretical and experimental research on the topic of adversarial machine learning as part of crafting project proposals for government R&D projects.

Sarasota, Florida Area

● Automated retrieval and processing of Call Detail Records from multiple sources using Python.● Preformed analysis on Call Detail Records using Python's sklearn package.● Produced visualizations of analysis using Python's matplotlib package.

Sarasota, Florida Area

Used Nvidia-Docker to construct a test suite that enabled experiments that used neural networks.Graded homework and held TA sessions for the following classes:● Calculus with Theory 1 & 2● Calculus 2● Real Analysis 1● Partial Differential Equations

Remote

● CodeU is an invite only mentorship program of 120 high-potential students designed to facilitate learning of software engineering skills.● As part of a remote team, designed, implemented, and tested key features such as search, login, and encrypted transmissions for a chat program written in Java.● Created and maintained a technical design document that detailed features to be added, their expected implementation details, and their eventual actual implementation.● Practiced the code review and version control processes using Git.● Regularly talked with our team’s Google host to discuss updates and trajectory.● Received strongly positive feedback; Flown out to attend CodeU day at Google’s Mountain View office.

Park City, Utah

● Invited and funded to participate in 3 week-long conference on Data Science run by the Institute for Advanced Study.● Attended sequences of lectures on statistics, machine learning, and wavelets, and many other lectures on various mathematical and data science topics.● Completed daily problem sets on data visualization in R and data processing with wavelets in Mathematica.● Regularly assisted fellow attendees in completion of problem sets, particularly on the subject of wavelets.

Albuquerque, New Mexico Area

● Independently added 2D Pyramid Recurrent Neural Network model, with associated datasets and demos into Cortex, an open source machine learning library built on top of Theano.● Reduced segmentation time from 20 minutes to 2 minutes per volume and improved error rate (.1 increase in Dice coefficient to .98) by implementing a 3D Up-Convolutional Neural Net into Keras. An author of the published result that was presented at the 2017 International Join Conference for Neural Networks.