Researching Vision Transformers (ViTs) with contrastive pretraining and layer-wise classifier evaluation, focusing on how representations evolve across encoder blocks, how feature separability can be quantified at different depths, and how multi-head training impacts accuracy and efficiency.

Developing an AI-powered agent using Crawl4AI to automatically discover, scrape, and track business-relevant articles, delivering structured, actionable insights in real time.

Explored core business and startup concepts using the Business Model Canvas. Researched product development, revenue, and impact models. Interviewed investors to understand funding and marketing strategies, and supported the growth of local tech startups through strategic insights and investment research.