🎓 B.Tech in Artificial Intelligence & Data Science 📍 Vishwakarma Institute of Technology (VIT), Pune 📊 CPI: 8.64

I'm an AI & Data Science undergraduate passionate about building intelligent applications using Machine Learning, Deep Learning, Generative AI, Large Language Models, RAG, NLP, Computer Vision, Data Analytics, and Backend Development.

I enjoy transforming real-world problems into practical AI-powered solutions, from automated invoice processing and document intelligence to multilingual data analytics and financial forecasting.

Currently, I'm exploring Agentic AI, advanced RAG systems, LLM applications, AI automation, scalable backend systems, and production-ready AI solutions.

• 🔭 Building AI/ML and Generative AI applications
• 🌱 Exploring Agentic AI, RAG, LLMs, AI Automation, and Cloud Technologies
• 🤖 Interested in Machine Learning, Deep Learning, NLP, Computer Vision, and Generative AI
• 📊 Passionate about Data Science, Data Analytics, ETL, and intelligent automation
• 💻 Experienced in developing Python-based AI and backend applications
• 🧠 Working with LLMs, embeddings, vector databases, document processing, and NLP
• 🏆 Best Pitch Award at Inspiron 4.0 National Hackathon among 300+ teams
• 🔬 Presented research on Optimized LSTM Networks for Robust Stock Price Forecasting at STAI 2026
• 👨‍🏫 Mentored 30+ students through the CII Eaton Changing Gears 7.0 initiative
• 📈 Always learning, building, experimenting, and improving

Python | Java | C | SQL | JavaScript | HTML | CSS

Machine Learning | Deep Learning | Data Science | NLP | Generative AI | LLMs | RAG | Computer Vision | OCR | EDA | Feature Engineering | Predictive Analytics

Llama 3 | Ollama | Google Gemini | BGE Embeddings | ChromaDB | Natural Language to SQL | Prompt Engineering | Semantic Search | Document QA

FastAPI | Flask | Streamlit | Pandas | NumPy | Scikit-learn | Matplotlib | Bootstrap

MySQL | MongoDB | ChromaDB | DBMS

Git | GitHub | VS Code | Google Colab | Power BI | Excel | Postman | ETL Pipelines | Data Visualization

Agentic AI | AI Agents | Advanced RAG | Cloud Deployment | Docker | CI/CD | Scalable AI Systems

Jul 2025 – Dec 2025

• Built LSTM-based Deep Learning models for financial forecasting using historical stock datasets.
• Improved prediction performance using SMA, RSI, MACD, and MinMax scaling.
• Performed model optimization using Random Search and dropout optimization.
• Evaluated models using RMSE, MAE, and R² metrics.
• Developed analytics dashboards and visualization applications using Flask and Streamlit.
• Worked on workflow automation, backend architecture, deployment workflows, and testing pipelines.

Tech Stack: Python • Flask • Gemini • MongoDB • OCR • Computer Vision • ETL

AI-powered accounts payable automation system designed to reduce manual invoice-processing effort.

• Built an automated invoice-processing workflow using OCR, document parsing, extraction, and validation.
• Implemented Computer Vision techniques for invoice segmentation and structured field extraction.
• Extracted vendor information, taxation details, invoice identifiers, and other structured data.
• Developed end-to-end ETL and document-processing pipelines.
• Reduced manual finance operations by approximately 80% through workflow automation.
• Built a scalable platform integrating OCR, validation, extraction, database storage, and analytics.

Tech Stack: FastAPI • BGE Embeddings • ChromaDB • Llama 3 • Python • RAG

A Retrieval-Augmented Generation system designed for intelligent insurance-document search and question answering.

• Built a complete RAG pipeline using BGE embeddings and ChromaDB.
• Implemented semantic search for retrieving relevant document information.
• Integrated Llama 3 for context-aware question answering.
• Automated processing of PDF, DOCX, and TXT documents.
• Implemented document extraction, chunking, metadata management, and indexing.
• Developed a FastAPI backend with structured JSON responses and confidence scoring.
• Designed the system to reduce hallucinations by grounding responses in retrieved document content.

Tech Stack: Python • Ollama • Llama 3 • MySQL • Streamlit • Pandas

A multilingual analytics assistant that allows users to interact with structured datasets using natural language.

• Developed an LLM-powered data analytics assistant for MySQL and CSV datasets.
• Implemented Natural Language → SQL conversion using Ollama Llama 3.
• Automated database querying and analytical workflows.
• Built interactive Streamlit dashboards with dynamic visualizations.
• Added automated chart generation for data-driven insights.
• Integrated file processing, MySQL connectivity, and multilingual support.

Tech Stack: Python • LSTM • Pandas • NumPy • Scikit-learn • Matplotlib

Deep Learning project focused on financial time-series forecasting.

• Built LSTM-based models for stock price prediction.
• Applied technical indicators including SMA, RSI, and MACD.
• Used MinMax scaling for sequential data preprocessing.
• Optimized hyperparameters using Random Search.
• Evaluated model performance using RMSE, MAE, and R².
• Presented research on Optimized LSTM Networks for Robust Stock Price Forecasting at STAI 2026.

Tech Stack: Python • FastAPI • BGE Embeddings • ChromaDB • Llama 3 • PyMuPDF

An intelligent document-processing and question-answering system for complex documents.

• Built semantic search using BGE embeddings and ChromaDB.
• Processed documents including PDF, DOCX, EML, and TXT.
• Implemented document chunking and metadata-aware retrieval.
• Integrated locally hosted Llama 3 for document-based question answering.
• Generated structured responses containing decisions, amounts, justification, and confidence.
• Designed the system around privacy-friendly local AI processing.

Tech Stack: Python • Transformers • PostgreSQL • Pandas • NLP

An NLP-based document processing pipeline for handling large collections of structured and unstructured data.

• Processed 100+ documents using embeddings and NLP techniques.
• Integrated document representations with PostgreSQL.
• Applied Transformer-based classification techniques.
• Built data-processing workflows for Excel and structured datasets.
• Combined NLP processing with database-driven analytics.

Tech Stack: HTML • CSS • JavaScript • Bootstrap • Node.js • PHP • JSON

A web-based real-estate platform developed as part of a VIT-sponsored project.

• Developed property listing and search functionality.
• Implemented filtering for property discovery.
• Worked with structured JSON-based property data.
• Built responsive interfaces using HTML, CSS, JavaScript, and Bootstrap.
• Included 50+ property listings in the application.

Tech Stack: Python • OpenCV • TensorFlow • CNN

Computer Vision application designed to recognize sign-language gestures.

• Used OpenCV for image processing.
• Built a CNN-based Deep Learning model.
• Processed visual gesture information for sign recognition.
• Explored the application of Computer Vision and Deep Learning for accessibility.

• 🥇 Best Pitch Award — Inspiron 4.0 National Level Hackathon among 300+ teams.
• 🔬 Presented research paper on Optimized LSTM Networks for Robust Stock Price Forecasting at STAI 2026.
• 📊 Earned IBM Certification in Data Analysis & Business Intelligence Fundamentals.
• 👨‍🏫 Mentored 30+ students through the CII Eaton Changing Gears 7.0 leadership development initiative.
• 💡 Participated in technical competitions and collaborative engineering projects.

• Artificial Intelligence
• Machine Learning
• Deep Learning
• Data Science
• Data Analytics
• Object-Oriented Programming with Java
• Web Development
• Database Management Systems
• Statistical Inference
• Computer Networks
• Operating Systems

• 🤖 Artificial Intelligence
• 🧠 Machine Learning
• 🧬 Deep Learning
• ✨ Generative AI
• 🧩 Agentic AI
• 🔤 Large Language Models
• 📚 Retrieval-Augmented Generation
• 👁️ Computer Vision
• 🗣️ Natural Language Processing
• 📊 Data Science & Analytics
• ⚙️ AI Automation
• 🗄️ Data Engineering
• 🔧 Backend Development

📧 Email: krishnanarkhede321@gmail.com

💼 LinkedIn: Krishna Narkhede

💻 GitHub: KrishnaNarkhede

📍 Pune, Maharashtra, India

"The best way to predict the future is to build it."

⭐ If you find my projects useful, consider giving them a star!

🚀 Keep Building. Keep Learning. Keep Innovating.