Erdi Kara - AI & ML Mentorship

Overview

Over the years, I’ve mentored 6 Master’s and 20+ undergraduate students on projects in machine learning and scientific computing. These mentorships were supported by grants from the DOE, NSF, Microsoft, Accenture, Morgan Stanley, and the Arthur P. Sloan Foundation.

We’ve worked on everything from computer vision, geospatial data, operator learning to spiking networks, and mathematical modeling in cancer immunology. Regardless, the goal was always practical: generating meaningful results and helping mentees grow into independent learners. Here are some of the stuff I worked on in this realm.

mama-tried.mp4

Microsoft - Funded Course: Deep Learning with PyTorch

This is Spelman College’s first deep learning course. We covered core tools like NumPy, Scikit-Learn, and PyTorch, along with real-world applications such as YOLO, GPT-3, Stable Diffusion, and Whisper. We collaborated closely with Microsoft data scientists to align the curriculum with industry needs. The students brought incredible energy to their final projects and completed pretty cool projects. Look below for some snapshots.

Mathematical Oncology: Modeling Cancer Immunotherapy

I created and taught a hands-on module on cancer immunology, where students explored how mathematical models can help us understand CAR-T therapies and tumor–immune interactions. We used real research papers as a foundation, then worked through simulations using ODEs, parameter fitting, and optimization. You can access their amazing presentations in our Github. See the figure on the right to appreciate what they did(:

Morgan Stanley Data Challenger

I led a team of five students to first place among fifteen teams in a two-stage competition judged by senior Morgan Stanley executives and ML experts. We combined geospatial analysis with semi-supervised learning to identify high-potential locations for business expansion with 98% accuracy. Along the way, we built the most comprehensive public datasets of U.S. zip codes, combining demographic and business metrics. Check out here to learn more. Github.