Flight Delay Prediction Model

● Developed a predictive machine learning model for an airline’s operations team to forecast flight delays based on historical flight, weather, and operational data. The project helped identify high-risk flights and improve scheduling decisions.

Smart Sales Advisor

My Role in Smart Sales Advisor: ● I led the end-to-end development of Smart Sales Advisor, an AI-powered sales analytics and forecasting platform for small business owners. My responsibilities included: ● Data Engineering & Analysis: Designed the data ingestion pipeline to handle CSV uploads, cleaned and transformed raw sales data, and implemented metrics like revenue growth, average order value (AOV), and top-selling products. ● Forecasting: Integrated Prophet to deliver 30-day sales predictions and visualize weekly, seasonal, and day-of-week trends. ● AI Insights: Used Hugging Face Transformers to generate actionable business insights and integrated Together.ai for an interactive AI sales advisor. ● Frontend Development: Built an intuitive Streamlit interface for real-time analytics, forecasting, and conversational AI. ● Deployment & Optimization: Deployed the app to Streamlit Cloud, optimized performance for real-time interaction, and incorporated feedback loops (thumbs up/down) to improve AI suggestions.

Electricity-Sales-Predictor

In this project, I was responsible for the end-to-end development of the Electricity Sales Predictor, including: ● Data Acquisition & Cleaning – Collected historical electricity sales data from the U.S. Energy Information Administration (EIA), handled missing values, formatted dates, encoded categorical features, and engineered relevant predictors. ● Exploratory Data Analysis (EDA) – Conducted detailed visual and statistical analysis to uncover sectoral trends, seasonal patterns, and price–sales relationships. ● Model Development – Built and tuned a Random Forest Regressor in Scikit-learn, achieving an R² score of 0.9917. Identified key predictive features such as customer count, pricing, sector, and month. ● Model Deployment – Serialized the trained model using Joblib, integrated it into a Streamlit web application with an interactive input form for real-time predictions, and deployed the app on Streamlit Community Cloud. ● UI/UX Design – Designed a clean, user-friendly interface to ensure accessibility for non-technical users, enabling instant predictions without manual code execution. ● Version Control & Collaboration – Managed source code, data files, and deployment scripts using GitHub for version control and project tracking. ● If you want, I can also make a shorter, CV-friendly version of this that’s more concise but still powerful.