Exploring pragmatic frontiers in machine intelligence—from multi-modal computer vision and computational biology to local, privacy-first AI agent frameworks.
Targeted exploration and building across high-impact AI subfields.
Leveraging local vision models and vector embeddings to downscale, tag, and semantically index large unstructured visual media collections.
Applying predictive structural models and machine learning pipelines to analyze molecular target structures and protein sequencing data.
Architecting localized multi-agent systems and file-based knowledge management setups for high-context, private, offline-capable assistance.
Bavi Work focuses on applied research that brings complex models out of theory and into useful, resilient workflows. We emphasize local-first architecture, privacy, data control, and targeted utility across scientific and personal productivity domains.