University of Utah · SCI Institute
Deekshith Dade
AI specialist focused on computer vision research, data-centric agentic systems, and streamlined product delivery for enterprise teams.
About
Over three years of building deep learning pipelines, leading advertising technology initiatives, and collaborating with research labs to translate algorithms into production software. I thrive in minimal, purposeful tooling and bring that mindset to every project.
3+
Years in research
10+
Projects shipped
Vision · Agents
Disciplines
Projects
Selected Work
Blog
Writing & Research Notes
MLX SAM3
A worklog to share some understanding of Meta's latest segmentation model SAM3 ported to mlx to run on Apple Silicon with a demonstrable application.
KV Cache in Large Language Models
A practical guide to understanding and implementing KV Cache in large language models. Implementation is using mlx framework, a NumPy-like array framework optimized for Apple Silicon's Unified Memory Architecture.
Images
Field Notes & Photography
