Portfolio - Hritik Rajan Bhatt

Hritik Rajan Bhatt

Data Aspirant

Resume

1

Machine Learning Project

1

Deep Learning Project

1

GenAI Project

About Me

Hello! I’m Hritik, a passionate Machine Learning and Data Science enthusiast who loves turning complex data into actionable insights and intelligent solutions. I enjoy exploring how data-driven systems can make real-world impact.

Tools and Technologies:
➡️ Proficient in Python, Scikit-learn, and Streamlit for building and deploying predictive models.
➡️ Experienced with Pandas, NumPy, and Matplotlib for data analysis and visualization.
➡️ Familiar with FastAPI for creating efficient, scalable APIs and integrating machine learning models into web applications.

Areas of Expertise:
➡️ Developing end-to-end machine learning pipelines — from preprocessing to deployment.
➡️ Building interactive data apps that make analytics accessible to everyone.
➡️ Applying statistical analysis and feature engineering to drive better predictions and decisions.

I believe in continuous learning and enjoy experimenting with new technologies in AI, data science, and automation. Currently, I’m enhancing my s

Key Skills

Python

NumPy

Pandas

Scikit-learn

Machine Learning

Data Analysis

Feature Engineering

Streamlit

FastAPI

EDA (Exploratory Data Analysis)

Jupyter Notebook

Data Visualization

Pytorch

My Projects

Credit Risk Analysis

Domain/Function: Machine Learning | Credit Risk Analysis | Financial Analytics

Car Damage Detection

Domain/Function: Deep Learning | Image Classification | Tranfer Learning | Model Evaluation

CityPulse: AI-Powered Civic Insights using RAG and Multi-Agent Systems

Domain/Function: Generative AI | NLP | RAG | Multi-Agent Systems

My Experience

Data Science Student (Fresher)

B.Tech – Data Science, Dwarkadas J. Sanghvi College of Engineering
Sep 2023 – Present

• Built and implemented end-to-end machine learning pipelines involving data preprocessing, feature engineering, model training, and evaluation using Python, scikit-learn, and PyTorch.
• Developed NLP and deep learning projects, including a multi-agent AI system for analyzing real-world data sources and generating actionable insights.
• Designed and deployed ML models using Streamlit and FastAPI to enable real-time predictions and user interaction.
• Applied statistical analysis and model evaluation techniques to improve prediction performance and ensure reliable outcomes.
• Gained hands-on experience working with tools like Pandas, NumPy, and Matplotlib for data analysis and visualization.
• Continuously learning and experimenting with AI, machine learning, and deployment techniques through practical projects.

Awards & Certificate

Python: Beginner to Advanced For Data Professionals

SQL for Data Science

Math and Statistics For AI, Data Science

Master Machine Learning for Data Science & AI: Beginner to Advanced

Deep Learning: Beginner to Advanced

Natural Language Processing

Let's Connect

Feel free to get in touch with me. I am always open to discussing new projects, creative ideas or opportunities to be part of your visions.

Download Resume

Resume

Mail Me

[email protected]

https://www.linkedin.com/in/hritik-bhatt-dj/

GitHub

https://github.com/Hritik123789/

Machine Learning Project

Credit Risk Analysis

Domain/Function: Machine Learning | Credit Risk Analysis | Financial Analytics

Built a machine learning model to predict credit risk using financial data. Implemented preprocessing, feature engineering, and handled class imbalance. Compared models like Logistic Regression, Random Forest, and XGBoost, achieving high recall, and deployed using Streamlit.

View Project
on GitHUB

View Project
on

Other Tools

Python

Numpy

Pandas

Scikit-learn

Matplotlib

Streamlit

The interface allows users to input financial and behavioral parameters like income, loan amount, tenure, and credit utilization. The system predicts default probability, credit score, and risk rating in real time, demonstrating end-to-end ML deployment.

Streamlit-based credit risk app where users input financial details like age, income, and loan metrics. The model predicts default probability (~7.47%), assigns a high credit score (855), and rates the borrower as “Excellent” in real time.

Project Details

I developed a credit risk prediction system using machine learning on structured financial data. The project involved end-to-end pipeline development including data preprocessing, feature encoding, scaling, and handling class imbalance to improve model reliability. Multiple models such as Logistic Regression, Random Forest, and XGBoost were trained and evaluated with a focus on recall to minimize false negatives in risk prediction. Performed detailed model comparison and error analysis using confusion matrix and performance metrics. The final model achieved high recall (~0.94) and was deployed using Streamlit to provide an interactive interface for real-time credit risk assessment and decision support.

Deep Learning Project

Car Damage Detection

Domain/Function: Deep Learning | Image Classification | Tranfer Learning | Model Evaluation

Built a car damage detection system using PyTorch and ResNet (transfer learning) to classify multiple damage types. Applied preprocessing, augmentation, and evaluation, achieving ~80% accuracy, and deployed via Streamlit for real-time image predictions.

View Project
on GitHUB

View Project
on

Other Tools

Python

Pytorch

Numpy

Pandas

Resnet

Torchvision

Image Augmentation

Streamlit

Matplotlib

Seaborn

Confusion Matrix

Streamlit-based UI for vehicle damage detection. Users upload an image via drag-and-drop, preview it, and get real-time predictions from a ResNet-based model. Designed for simple, fast, and user-friendly interaction.

Displays uploaded car image with visible rear damage and the model’s prediction “Rear Breakage.” Demonstrates real-time inference capability using a trained deep learning model for accurate vehicle damage classification.

Project Details

Developed an end-to-end Vehicle Damage Detection system using deep learning and computer vision techniques to automatically classify car damage from images. Leveraged PyTorch with transfer learning (ResNet) to build a robust image classification model capable of identifying multiple damage categories such as dents, scratches, and rear breakage. Implemented data preprocessing and augmentation pipelines to improve model generalization and handled class imbalance to enhance performance. Achieved around 80% validation accuracy and conducted detailed evaluation using confusion matrices and class-wise metrics. Deployed the solution using Streamlit, enabling users to upload images and receive real-time predictions through an interactive web interface. The system demonstrates practical application of AI in insurance assessment and automated vehicle inspection workflows.

GenAI Project

CityPulse: AI-Powered Civic Insights using RAG and Multi-Agent Systems

Domain/Function: Generative AI | NLP | RAG | Multi-Agent Systems

Built a RAG-based multi-agent AI system using AWS Bedrock to analyze data from news, Reddit, and public sources. Implemented NLP, embeddings, and web scraping to generate real-time civic insights, and enabled natural language querying via a voice-enabled interface.

View Project
on GitHUB

View Project
on

Other Tools

Python

AWS Bedrock

Amazon Nova

Amazon Titan Embeddings

LangChain

Selenium

Fast API

Pandas

Numpy

AI-powered dashboard showing real-time civic insights for Mumbai, including alerts, permits, and topic trends. Aggregates multi-source data and presents key metrics for quick understanding and decision-making.

Displays AI-generated civic alerts with categorization, scoring, and timestamps. Demonstrates real-time analysis of news and data sources to highlight critical safety and infrastructure updates.

Project Details

Built a Generative AI-based platform to analyze and generate real-time civic insights from multi-source data including news APIs, Reddit, and government permit databases. Implemented Retrieval-Augmented Generation (RAG) using AWS Bedrock (Amazon Nova & Titan Embeddings) to process structured and unstructured data. Developed multi-agent workflows for sentiment analysis, topic clustering, and trend detection. The system provides features like AI-generated daily briefings, real-time alerts, permit tracking, and community insights dashboards. Integrated web scraping (Selenium) and automated pipelines for continuous data collection. Also enabled natural language interaction through a voice-based assistant. Designed for scalable, real-world applications in urban monitoring and decision-making.