Ibrahim Saeed Email & Phone Number

Toronto, On, Ca

New York, United States

Leading Purdue Team that works with Google’s Tensorflow Model Garden Team on experiments to build and improve low-power computer vision models YOLOX-Tiny, YOLOX-nano and MobileNetV3.

Improved Graph Link Prediction model SEAL speed performance by 70% in Pytorch with ~0.29 accuracy (HITS@20)on the OGB-ddi dataset.• Experimented using different machine learning models, papers and techniques (k-hop neighbours, adamic adar,labelling) and wrote 5 reports detailing how I improved SEAL.• Technology: Pytorch, Python, cProfile, Jupyter Notebooks

Mountain View, California, United States

Redesigned and Built a machine learning privacy pipeline from scratch using C++ and Flume, used for processing terabytes of training data.• Completed entire development process for constructing a pipeline, including writing a design document,under-going design review, writing unit/integration tests, going though code reviews and launching internally.• Deepened knowledge in parallelism with Flume, best practices for designing and writing integration tests for large-scale pipelines.• Technology: C++, Flume

United States

• Utilized ChartJs to show Growth Degree graphs using up-to-date weather data• Wrote Dexie queries and manipulations to manage farmer’s field data in indexedDB• Designed and created forms and reusable components in Svelte-kit

• Developed Python script and Bash Shell script to filter out irrelevant features which lead to a ~99% reduction in the resulting file’s size, ~99% reduction in irrelevant data and improved process to save 5 hours a week.• Constructed data-processing pipeline in Flume (Java) to convert between different Google Maps Protobufs• Completed entire development process, including writing two design docs and usage documentation, implementation, writing unit tests (Junit4 and Mockito), going through code reviews and launching internally.• Eliminated technical debt by solving bugs during Bug Bash and implementing two feature requests for user.

Created a responsive full stack AI-powered (GPT-3) data analytics startup recognized, mentored and funded by Microsoft For Startups that had over 10 active companies/clients during BETA access.• Implemented 60% of UI (Forms, Dashboards, Graphs) with React and Tailwind CSS.• Built backend endpoints using Flask and NoSQL database Firestore.