class Paras:

    name       = "Paras Khutwad"
    alias      = "paras-the-coder"
    status     = "B.Sc. Statistics Student @ Modern College, Pune"
    role       = "Aspiring AI Engineer | Builder | Learner"

    currently  = [
        "Building Agentic AI & RAG systems",
        "Exploring LangGraph & multi-agent pipelines",
        "Studying Machine Learning & Deep Learning",
        "Developing Computer Vision projects",
        "Seeking remote AI/ML internship opportunities",
    ]

    interests  = [
        "Agentic AI",
        "Retrieval-Augmented Generation (RAG)",
        "Fraud Detection & Predictive Analytics",
        "Computer Vision & Deep Learning",
        "Generative AI & LLMs",
    ]

** Agentic Hybrid RAG Assistant** — My most technically differentiated project

An agentic retrieval system built with LangGraph that goes beyond standard RAG. Features a self-critique loop that evaluates its own answers before returning them, hybrid retrieval (dense + sparse), Pinecone vector storage, Tavily live web-search fallback, Groq Llama-3.3-70B inference, and a FastAPI + SSE streaming backend. Supports multiple documents with persistent memory across turns.

LangGraph LangChain Pinecone Groq FastAPI Tavily HuggingFace SSE Streaming

** FraudGuard AI** — Insurance fraud detection with ML explainability

A fraud detection system trained on insurance claims data. Implements XGBoost and Logistic Regression with SMOTE for class imbalance, real SHAP explainability to surface which features drive each prediction, a risk scoring engine, and an interactive Streamlit dashboard.

XGBoost Scikit-Learn SMOTE SHAP Pandas NumPy Streamlit Feature Engineering

** Solar Panel Defect Classification** — Computer Vision for clean energy

A CNN-based image classification system that detects defects in solar panels using transfer learning with ResNet50. Built on TensorFlow/Keras, preprocessed with OpenCV, deployed live on Streamlit Cloud.

ResNet50 TensorFlow Keras OpenCV Transfer Learning Streamlit Image Classification

Currently Learning:
  - LangGraph & Agentic AI
  - Retrieval-Augmented Generation
  - Machine Learning
  - Deep Learning & Computer Vision
  - FastAPI & AI Backend Development
  - Prompt Engineering

Current Goal:
  - Secure an AI/ML Internship
  - Improve Machine Learning Skills
  - Build Production-Ready AI Systems
  - Contribute to Open Source Projects

Currently seeking: Remote AI/ML Internships  •  Open to project collaborations

 

  

💬 Open to: AI/ML Internship Opportunities  •  Project Collaborations  •  Learning Discussions

Building one model at a time — from statistics to intelligence.