AI & Data Science Portfolio

G R
Muthamil
harasi

Turning raw data into real-world decisions. Final-year AI & Data Science student specializing in analytics, dashboarding, and machine learning.

View My Work Get in Touch
Scroll
01 About

The person
behind the data

Final-year AI & Data Science student with hands-on experience in data analysis, dashboarding, and machine learning. I specialize in transforming messy, scattered data into clear, actionable insights that drive real decisions.

From cleaning 20,000+ healthcare records to building interactive Tableau dashboards, I approach every problem with both technical rigour and business thinking — because data only matters when it changes something.

20K+
Records processed
4
End-to-end projects
15+
Data fields analyzed
3
Tools mastered
02 Value

Why work
with me

01

Real Data Experience

Worked with 20,000+ records and complex real-world datasets, not just toy examples.

02

End-to-End Projects

From raw cleaning → EDA → analysis → dashboard → storytelling with insights.

03

Business Thinking

Focus on decisions and outcomes, not just models. Data should move the needle.

04

Fast Learner

Quickly adapt to new tools, domains, and team needs with minimal ramp-up time.

03 Skills

Capabilities

Data Analysis & EDA→
Machine Learning→
Tableau Dashboard Development→
Power BI Reporting→
Python, SQL, Advanced ExcelAnalytics→
Data Cleaning & Transformation→
Business Intelligence→
Prompt Engineering→
04 Work

Selected
projects

03
AI-Powered Stock Market Prediction & Analysis
Python Flask FinBERT NLP Machine Learning
Problem Difficulty in extracting meaningful investment insights from large volumes of real-time financial news and market data.
Approach Built an AI-driven financial analytics platform using Flask, FinBERT-based sentiment analysis, and real-time news APIs to analyze market sentiment and stock trends.
Impact Enabled real-time financial sentiment monitoring, improved market insight generation, and supported data-driven investment analysis through interactive analytics dashboards.
View Code
02
Bank Loan Financial Analytics
Python EDA
Problem Banks lacked a unified view to identify high-risk loan segments, default patterns, and NPA exposure across customer demographics.
Approach
  • Cleaned 10,000+ loan records using Pandas
  • Created risk categories, DTI ratio, and default flags
  • EDA Analysis
Insights
  • Very High Risk borrowers default at 3× the rate of Low Risk
  • Debt consolidation accounts for 40%+ of all loans
  • NPA exposure concentrated in large loan category
View Code
Model
Loan Risk Prediction System and Analytics
Python Machine Learning Streamlit Pandas Scikit-learn
Problem Difficulty in accurately identifying high-risk loan applicants and analyzing financial risk patterns from customer financial data.
Approach Developed a Machine Learning-based loan risk prediction system using data preprocessing, feature engineering, and predictive modeling integrated with an interactive Streamlit dashboard.
Impact Improved credit risk assessment through predictive analytics, enabled real-time loan risk prediction, and supported data-driven financial decision-making.
View Code
03
E-Commerce Analytics Using SQL
SQL Joins Window Functions CTEs
Problem An e-commerce company needed visibility into revenue trends, product performance, customer behavior, and profit margins across cities and categories.
Approach
  • Designed a 4-table relational database
  • Wrote 18 SQL queries from basic to advanced
  • Built reusable Views and automated monthly report procedure
Insights
  • Electronics drives 68% of total revenue
  • Tamil Nadu is the top state by order volume
  • Champions segment customers spend 4× more than At Risk
View Code
04
Automobile Sales Dashboard
Power BI Power Query DAX
Problem Analyze revenue trends across products and regions with no visual overview.
Approach Data cleaning via Power Query, DAX KPIs, and an interactive Power BI dashboard with cross-filter capabilities.
Impact Identified top product line (~39% of revenue) and highest-performing country (USA), enabling strategic resource allocation.
View Code
05
Healthcare Patient Analytics
Python Tableau Calculated Fields Dashboard Design
Problem Healthcare organizations lacked a unified view to identify which conditions, age groups, and hospitals needed the most attention.
Approach Analyzed by using Python(Pandas),ETL and EDA ,Built an interactive Tableau dashboard with KPI cards, bar charts, line charts, and global filters using calculated fields across 15+ data fields.
Impact Revealed young patients are most affected by obesity; arthritis has the highest patient count — helping hospitals plan preventive care more effectively.
View Code View Dashboard
05
Sales Performance Analysis
Python EDA
Problem Identify trends and revenue gaps across 50,000+ sales records with no structured view.
Approach Python EDA to surface key performance patterns and segment revenue by region and product.
Impact Revealed performance gaps and quantifiable growth opportunities reviewed directly by leadership.
View Code
06
Space-Based Data Center Cooling Analysis | Selected – MoE, AICTE, Wadhwani Bootcamp
Thermal Simulation Ansys Woekbench Radiator, Connector, Silicon Chips Heat Analysis
Problem Improve data center cooling efficiency by evaluating heat dissipation strategies in space environments.
Approach Thermal simulation and heat dissipation modeling to evaluate space-based cooling approaches.
Impact Identified key optimization opportunities for energy-efficient data center operations.
View Code View Model
05 Experience

Where I've
worked

Data Science Intern
TechCiti Software Consulting Pvt. Ltd. — Bengaluru, India
May 2025
June 2025
→Processed, validated, and quality-checked 20,000+ healthcare records, significantly improving data accuracy and reliability.
→Conducted Exploratory Data Analysis (EDA) to identify trends, anomalies, and inconsistencies across structured datasets.
→Performed end-to-end data cleaning and transformation pipelines for structured business reporting.
→Collaborated with cross-functional teams to deliver data-driven insights that informed key business decisions.
06 AI Demo

AI project
in action

Mental Health Chatbot

A conversational AI application built for mental wellness support using Amazon PartyRock. Designed to provide empathetic, structured responses to users navigating emotional challenges — accessible, private, and always available.

Open Live App →
Mental Health Chatbot — UAlive on Amazon PartyRock
07 Contact
Let's build something with data

Open to internships, collaborations, and full-time data roles. I respond within 24 hours.

Email
muthamilharasigr@gmail.com
↗
LinkedIn
linkedin.com/in/muthamilharasigr
↗
GitHub
github.com/Muthamilharasi
↗