Implementing full-stack core features for a mom-mentorship platform leveraging Node, React, PostgreSQL, JavaScript, Docker. Currently working on a refined search functionality with both syntactic and semantic matching. Features are shipped to production quickly & impact real customers.

Spring 2026

Expanding MPICH-RCCL integration to exascale testing on the supercomputer Frontier (AMD GPUs) at Oak Ridge National Laboratory, evaluating additional collectives, and applying the work to real-world scientific applications. This research advances GPU-accelerated MPI for distributed AI and HPC workloads, producing broadly publishable results that support Argonne’s mission in high-performance computing.

As part of the MPI4AI initiative at Argonne National Lab, I worked on integrating AMD’s RCCL into MPICH, a leading open-source MPI implementation. The goal was to enhance MPI’s performance for AI workloads by leveraging RCCL’s GPU-optimized collective operations within the MPI framework. I implemented and tested the Allreduce backend, evaluated performance on systems like JLSE and Frontier, and explored hybrid approaches that mix RCCL with native MPI to deliver optimal results. This work contributes to a broader effort to unify communication libraries across HPC and AI platforms.

Supervisor: Dr. Mike Wilkins, Maria Goeppert Mayer Fellow

Implemented full-stack authenticated Microsoft sign-on using Django and React with secure session handling on Azure. Automated internal security and access-control workflows by integrating Git CI to reduce manual overhead. Performed prompt engineering tasks for refined LLM use.