Project Demos
Watch live demonstrations of my key projects. Click on any demo to see the project in action.
AI Customer Support Ticket Resolver
Automated customer support using AI agents and MCP
Generative AI Travel Chatbot
AI-powered travel assistant with web scraping
Coming Soon
Fashion Recommendation System
AI-powered fashion recommendations with image analysis
About Me
Data Scientist with 3.9 years of experience developing full-stack data science projects across the insurance, cyber security and e-Commerce domains, leveraging machine learning, deep learning, Gen AI and MLOps to solve complex business challanges.
Technical Skills
Programming & Data Analysis
Python
SQL
PySpark
Pandas
NumPy
Matplotlib
Seaborn
Statistics
Hypothesis Testing
Probability
Statistical Analysis
A/B Testing
Machine Learning
Scikit Learn
Linear Regression
Logistic Regression
Ridge Regression
Decision Trees
Random Forest
SVM
XGBoost
k-NN
Naive Bayes
K-Means Clustering
Deep Learning
Neural Networks
CNN
RNN
LSTM
Transformers
TensorFlow
PyTorch
Natural Language Processing
word2vec
BERT
GPT-based models
Fine-tuning
Text Classification
Named Entity Recognition
Generative AI
LLM
LangChain
LangGraph
AI Agents
Multi Models
RAG Databases
Vector Embeddings
Cloud & DevOps
Amazon SageMaker
Amazon Bedrock
Azure DevOps
Azure AI
CI/CD
MLOps
Git
DVC
Dagshub
MLFLOW
Docker
Airflow
Work Experience
Associate Data Scientist
Versatile Commerce
Nov 2021 – Feb 2025
Data Science Intern
iNeuron
Nov 2020 – Mar 2021
My Certifications
My Projects
AI Customer Support Ticket Resolver Using MCP
- This Project uses large language models to automate customer support. It classifies tickets, analyzes content, generate and send responses automatically to the given customer email address.
- Built with Streamlit and MCP (Model Context Protocol) Inspector Tool.
- Accepts customer support messages or queries and uses AI to understand the issue and generate a helpful reply.
- Detects urgency and classifies the type of request, automatically sends responses via email.
- Automatically logs tickets into a Google Sheet and provides a simple Streamlit web interface with MCP Inspector Tool.
LLM
MCP
Streamlit
AI Agents
Google Sheets API
Email Automation
Generative AI-Powered Travel Chatbot
- Developed a Generative AI-powered travel chatbot for my travel website, designed to retrieve and present information dynamically using web scraping and conversational AI.
- The chatbot delivers an interactive user experience by engaging in real-time conversations.
- It features AI agents that handle cases where information is unavailable on the website.
- Upon user consent, these agents fetch the required data from external sources, ensuring a seamless and user-controlled interaction.
LangChain
Chroma DB
FastAPI
Groq
Human-in-the-Loop AI Sales Agent
- Designed an AI-powered sales team with n8n, integrating human-in-the-loop functionality for personalized email outreach.
- Created a workflow where the AI Sales Agent generates a personalized email for review.
- Implemented a continuous feedback loop where a human reviews, provides feedback, and the AI Revision Agent adjusts the email until it's approved and ready to send.
n8n
AI
Automation
Email
Cache Augmented Generation (CAG) Chatbot
- Developed a professional chatbot that reduces response time and improves performance using smart caching mechanisms.
- Integrated custom vector embeddings, Python subprocess-based LLM querying, and Mistral-7B-Instruct-v0.3.
- Designed an intuitive front-end using Streamlit for seamless interaction.
Streamlit
Mistral-7B
Vector Embeddings
LLM
Fashion Product Recommendation System
- Built a content-based recommender system with outfit compatibility and image-based recommendations.
- Implemented TF-IDF for text-based product search, fine-tuned ResNet50 for image recommendations, and optimized FAISS for fast similarity search.
- Deployed on Streamlit with Google Gemini API for real-time outfit analysis.
TF-IDF
ResNet50
Google Gemini API
FAISS
Stock Market AgenticAI
- Developed an AI-driven stock market assistant using phidata for real-time financial insights.
- Leveraged YFinanceTools, Google Search, and DuckDuckGo for live stock price analysis and updates.
- Built a multi-agent system for structured decision-making and stock recommendations.
phidata
YFinanceTools
DuckDuckGo
Multi-agent
Automated Predictive Maintenance Pipeline for APS
- Developed an automated pipeline to predict failures in heavy-duty vehicle APS by distinguishing component failures from other vehicle issues.
- Utilized Apache Kafka for data ingestion, transferring sensor data to MongoDB, and trained with various machine learning models (Random Forest, XGBoost) for accurate failure predictions.
- Implemented a scalable End-to-End pipeline with CI/CD using GitHub Actions, deploying on EC2 and containerizing with Docker.
Apache Kafka
MongoDB
Docker
EC2
XGBoost
Book Recommender System
- This project demonstrates a Book Recommender System utilizing Collaborative Filtering to recommend books to users based on their preferences.
- The system helps users find books they are likely to enjoy, based on the choices of others with similar tastes.
- It trained on a collaborative filtering model on the dataset using user ratings.
Collaborative Filtering
Recommendations
Python
IPL Score Prediction
- Developed a machine learning model to predict IPL match scores using Linear Regression and Ridge Regression.
- Performed data preprocessing and model evaluation to ensure optimal performance.
- Utilized Flask to deploy the model in a web application for real-time predictions.
Linear Regression
Ridge Regression
Flask
Python