Led the inference team to reduce latency and price using heterogenous computing and computer architecture analysis (plus kernel development).

Created and packaged a distributed task queue for asynchronous processing with atomic retrieval and detailed logging.

Fine-tuned + deployed SLMs on classification and generation tasks, implementing a continuous training pipeline.

Developed a client for an internal, modular MCP system supporting multi tenancy, authentication, and authorization.

Created fuzzy search engine to match Salesforce cases to accounts, and later added a vector database to improve accuracy + allow for gradual improvement.

Developed a course project which tested students' ability to develop a shallow water equation solver in parallel using GPUs and multiple computer cores using CUDA and MPI.

Held weekly office hours where I walked students through developing optimized matrix multiplication kernels, simulating water particle dynamics in a dam scenario, and completing the above project.

Served as a community leader and peer educator, encouraging residents to respect one another and get involved with floor, hall and area activities, and provided information about campus resources.

Made rounds of the community to meet residents and look for violations of policy, and served on-call for 24 hour periods where residents could call and ask questions or resolve issues.

Managed distributed computing systems and user accounts, answered researcher queries, kept records and generated progress reports, and assisted with workshops.

Completed a project to autonomously search for optimal neural network designs and published a research paper on my work as part of the High Performance Computing Modernization Program (HPCMP) Internship Program (HIP).