Anushi User

I'm building practical AI systems across computer vision, LLM agents, and backend APIs. At Escrow Infotech, I developed an AI-based conflict resolution system that combined ML inference pipelines, Node.js REST APIs, and the Gemini framework.

My work includes text analysis models that classify conflict types from conversational signals and generate context-appropriate resolution strategies. I also implemented scalable backend architecture for low-latency responses and enterprise workflow integration.

In my agricultural automation project, I built a hierarchical deep learning pipeline for fruit disease detection, reaching 98.7% classification accuracy across five mango disease types under controlled conditions. I processed and augmented 5,000+ agricultural images, used U2-Net for saliency-based background removal, and applied ResNet-50 transfer learning for disease classification.

I've also built an AI SQL agent with LangGraph and FastAPI, a RAG codebase assistant using ChromaDB and Groq, and a safety-focused navigation system using DBSCAN clustering and Dijkstra's algorithm. Earlier, I tested franchise management software at Sonani Foundations, documenting bugs and supporting quality assurance.