André Evard Email & Phone Number

Boston, Massachusetts, United States

Lemont, Illinois, United States

Adapted Megatron DeepSpeed, HuggingFace, and other Vision Transformers for in-house supercomputers.Conducted and automated 200+ hour-long ViT experiments to gather hardware and model performance metrics.Informed the strategic differences between architectures for long-term development.

Boston, Massachusetts, United States

Lemont, Illinois, United States

Visualized the efficacy of various UNet models with ALCF's AI-Testbed team, running these image recognition models on an assortment of super computers, and recorded the costs associated in running them. Power, runtime, memory usage, card utilization, and parallelization were some primary metrics I analyzed. Tested the architecture of various in-development super computers as part of performing model comparisons, including machines built by SambaNova, Intel Habana, and a more stable XC40 by Cray. Carefully considered a wide array of professional-grade models before deciding to run the UNet. I also considered orchestrating a T5 transformer or other NLP-centric models, we decided image recognition took priority.

Greater Boston Area

Tutored a wide array of students in office hours and graded homework on a weekly basis. I also helped to proctor and grade tests, and held recitations.Taught an assortment of classes through the years, including three semesters of Data Science and Algorithms, two of Operating Systems, and one in Structure and Interpretation of Computer Programs.

Lemont, Illinois, United States

Assisted in development on the open-source scheduler, OpenPBS, which Argonne's ALCF software support teams ensure worker properly on their particular high-scale computer architecture. On that team, I wrote a Python application that parses C code to instrument it with runtime tracking data.Generated in-depth runtime data, then cleaned and visualized it to point the senior developers toward bugs that were causing compounding slowdowns in the scheduler, helping our researchers utilize the their models sooner and faster.

Greater Boston Area

Integrated an open source Java machine learning library into TIBCO's StreamBase visual language platform. Using a variety of online (streaming) learning techniques to ensure relative usage uniformity, I orchestrated a number of models including KNN, K-Means, LogReg, and Ada-Boost.Demonstrated the app's applicability with a few sample programs, all of which I wrote with thorough documentation, and included as demos with an installation of the app.

Greater Boston Area

Established monitors for Tamr's DevOps team to maintain their virtual machines with, renewed licenses and ensured they were all up-to-date, and kept the remote machines up to date using an assortment of tools. I also helped solve other engineers' network, cloud, and related issues while on call.Developed a whole web-application project as the internship's culmination, written using Python and Javascript. Via the app I enabled the whole development team to monitor virtual machines on the Google Cloud Environment, and coordinated the company's essential internal websites.

Master Of Science - Ms, Data Science

Bachelor'S Degree, Mathematics And Computer Science