At Credes Labs, I developed and optimized Flutter modules in a live production codebase. I improved maintainability with reusable widgets and reduced UI render times.
I built Krishivani, an AI-powered plant disease and market intelligence platform combining image classification, multimodal question answering, knowledge retrieval, and market-price forecasting. Its EfficientNetB0 model was trained on 54,000+ images across 38 disease classes and achieved 98% validation accuracy.
For Krishivani, I also integrated Qwen2.5-VL 3B with semantic retrieval and developed FastAPI inference services, using Supabase Row-Level Security to protect application data.
I developed LoopedIn, a cross-platform second-hand fashion marketplace, with a content-based recommendation engine and a PostgreSQL/Supabase backend. My work also includes solving 700+ LeetCode problems and heading AKGEC's Cultural Society, where I led a 40-member team organizing institutional events.
