Building Geospatial AI Agents and internal tooling for Matter’s Geospatial Intelligence Platform (map automation + agent tools/workflows).

Course Assistant for CS-UY 2413 Design and Analysis of Algorithms working with Professor Boris Aronov

Course Assistant for CS-UY 2413 Design and Analysis of Algorithms working with Professor Lisa Hellerstein

Assisting in Prof. Cem Deniz’s lab, focusing on multimodal deep learning for knee osteoarthritis prognosis using medical imaging and AI methods.

Patent-pending model optimization work from my AMD internship, where I boosted inference throughput by 45% across 15+ GPU/APU workloads by removing kernel bottlenecks and deploying at scale with Docker/Helm on Kubernetes.

Implemented BF16/BFP16 quantization (converting Microsoft A16W8 models across 100+ ops with 97.1% YOLOv3 accuracy), integrated PyTorch Dynamo + torch.fx for PTQ/mixed-precision LLMs/CNNs/Transformers, and used Bayesian/RL search to auto-tune per-layer bit-widths and scaling for real-time inference.