At Yash Technologies, I built NutriScanAI, a personalized nutrition assistant using a five-agent LangGraph pipeline. I integrated Groq LLM for health scoring and dietary insights, and connected product data sources with an LLM fallback.
I deployed NutriScanAI as a Streamlit app with user authentication, scan history, and exportable reports. At Planto.ai, I tested the Image-To-Code feature, contributing to a 10% performance improvement, and researched potential B2B clients and markets.
At ZingMind Technologies, I implemented Llama2-7B fine-tuning on an internal domain corpus using Python and the Hugging Face stack, and applied prompt engineering to improve LLM responses.
In my projects, I’ve developed agentic AI systems for security analysis and AI service orchestration, along with a research paper summarizer. I’ve used LangGraph, RAG, and FAISS to build systems that analyze reports, recommend remediation steps, and produce structured summaries.
