This 4-page Performance Dashboard was designed to deliver a modern, interactive view of key business metrics across products, locations, and sales channels.

I focused on creating a visually appealing, modern UI using clean layouts and custom SVG elements for an enhanced data storytelling experience.

The project required writing complex DAX measures to calculate dynamic KPIs, time-based comparisons, and multi-dimensional aggregations.

Through this build, I deepened my understanding of data modeling, UI consistency, and performance optimization combining technical precision with modern design principles.

Tools: MS Excel, SQL, PowerBI

Built a Databricks-powered big data pipeline to process and analyze over 480K global clinical trials using PySpark (RDDs, DataFrames, and SQL). Automated data cleaning, transformation, and integration into Power BI to visualize key insights on study trends, sponsors, and completion rates.

This project strengthened my skills in big data processing, ETL automation, and interactive dashboard design for large-scale analytics.

Tech Stack: Databricks, PySpark (RDDs & DataFrames), SQL, Power BI

I designed an interactive Medical Claims Dashboard in Tableau to help visualize key performance metrics such as total claims, approval rates, and monthly claim trends. The dashboard allows users to monitor claim status distribution, analyze approval patterns by CPT code, and identify high-performing or delayed claim areas. My focus was not only on delivering insight but also on creating an intuitive, aesthetically pleasing design that simplifies complex data for non-technical users.
This project significantly improved my data visualization and UI/UX design skills. I deepened my understanding of color theory, layout composition, and accessibility, leveraging tools like Adobe Color, Coolors, and Color Hunt to create a cohesive and balanced color palette. It also enhanced my ability to combine domain knowledge, analytical storytelling with design aesthetics to make dashboards both functional and visually engaging.

Tech Stack: Tableau, Tableau Prep, Microsoft Excel

Unemployment in Africa – A Data-Driven Hackathon Project

Developed during the 10Analytics Hackathon, this project explores the root causes and socio-economic factors driving unemployment across Africa using data visualization and analysis in Power BI.

The dashboard examines the relationship between unemployment rates, gender, education expenditure, business density, and access to electricity, providing actionable insights and policy-driven recommendations.

Tech Stack: PowerBI, Excel, Data modelling, DAX 

This personal project began with curiosity — I wanted to explore my Netflix viewing habits, highlight my favorite shows, most active viewing times, and watch patterns among friends sharing the account.

I extracted data from my Netflix app and built an end-to-end ETL pipeline:

• Data cleaning & modeling: Processed and transformed raw files using Tableau Prep to create structured datasets.

• Visualization: Designed an interactive Tableau dashboard to analyze watch activity by time, genre, and day of the week,

• Key insights: Discovered Friday as my peak viewing day, with evening sessions dominating my watch patterns.

What I Learned:
This project deepened my understanding of ETL workflows, data cleaning, and data visualization using Tableau Prep and Tableau Desktop reinforcing how personal curiosity can evolve into meaningful analytics.

Tech Stack: Tableau, Tableau Prep, Google sheet, Data modelling  

I designed a Marketing Performance Dashboard in Tableau to monitor and analyze campaign effectiveness across key metrics such as spend, impressions, conversions, and ROI. The dashboard provides a clear view of marketing performance over time, breaking down data by channel and campaign to reveal which strategies drive the best results.

What I Learned:
This project deepened my understanding of marketing analytics, KPI tracking, and storytelling with data. I improved my ability to align visual insights with business goals, ensuring that every chart serves a strategic purpose. I also refined my Tableau dashboard design skills, emphasizing simplicity, interactivity, and performance optimization for decision-makers.

Tech Stack: Tableau · Excel · Data Cleaning & Transformation · KPI Analytics

This project focused on extracting and analyzing key themes across a large corpus of research abstracts (text data) related to "Artificial Intelligence and Machine Learning" using Natural Language Processing (NLP) and Topic Modeling.

Key Insights

Discovered 7 dominant topics, including machine learning, deep learning, neural networks, cancer prediction, and data systems.

Revealed the dominance of AI-related research trends, with strong overlaps between deep learning and medical applications.

What I Learned

Advanced NLP workflows: text preprocessing, vectorization, and embedding generation.

Tech Stack: Python, NLTK, spaCy, scikit-learn, Gensim, BERTopic, Matplotlib, Plotly, WordCloud

I built a real-time bicycle rider detection system using advanced AI models (YOLOv7 and YOLOv8) to enhance cyclist safety and support smarter traffic management. 

I personally collected and labeled over 170 images and videos of cyclists across different lighting and weather conditions, preparing the data through cleaning and augmentation before training both models. While YOLOv7 performed well, YOLOv8 delivered higher precision, faster processing, and better adaptability for real-world use. 

Through this project, I learned how high-quality data, automation, and model optimization come together to make AI systems both practical and impactful.

Tech Stack: Python · YOLOv7 · YOLOv8 · TensorFlow · PyTorch · Roboflow · Google Colab · OpenCV

I developed a customer segmentation model for a bank’s telemarketing campaign using K-Means and Hierarchical Clustering. The goal was to group customers based on demographics, financial behavior, and campaign interactions to improve targeting efficiency. 

After cleaning and standardizing the dataset, I applied K-Means (k=3) — achieving a silhouette score of 0.53 — which identified three distinct groups: high-, middle-, and low-income customers. Hierarchical clustering (silhouette score: 0.47) provided additional insights into relationships between these groups. 

This project strengthened my understanding of EDA, feature scaling, and unsupervised learning, proving how data-driven segmentation can make marketing strategies more precise and cost-effective.

Tech Stack: Python · Scikit-learn · Pandas · Matplotlib · Seaborn · K-Means · Hierarchical Clustering

I built a machine learning model to predict student dropout and academic success helping schools identify at-risk students early and improve retention. Using a dataset of 4,400+ students, I analyzed demographics, academic performance, and financial factors to uncover patterns that influence student outcomes. 

I applied KNN and Decision Tree models, optimizing them with GridSearchCV and balancing the dataset using SMOTE. The KNN model achieved 71.4% accuracy, outperforming Decision Tree and proving more effective in identifying students likely to drop out. 

This project deepened my understanding of data preprocessing, feature selection, and model optimization, showing how predictive analytics can support better educational strategies.

Tech Stack: Python · Scikit-learn · Pandas · NumPy · Matplotlib · Seaborn · SMOTE