My Projects

"some of the projects I've worked on that highlight the skills I've learned, the tools I've used, and how I solve real problems through creative and practical solutions."

Academic Projects

ChatBot

Developed an AI Chatbot for Financial Analysis, offering financial information through PDF document analysis and live stock market data. It uses the llama3.2 model for language understanding via OllamaLLM, PyPDF2 for PDF processing, FAISS for document search, and integrates with Finnhub API and yfinance for real-time and historical stock data. Made a streamlit based UI for interactive user queries and visualization.

GitHub Link

Concall Sentiment Analysis

Applied Natural Language Processing (NLP) to analyze quarterly earnings call transcripts of Neuland Laboratories, a pharmaceutical company. Extracted sentiment and other textual features from these calls and investigate their potential correlation with stock price movements. Utilized libraries like vaderSentiment, afinn, PyPDF2, and nltk for sentiment analysis and text processing. Loading metadata and stock prices from an Excel file, then extracting text content, page counts, and word counts from PDF transcripts using PyPDF2. Performed sentiment scoring using VADER to quantify the emotional tone. Uncovering hidden patterns and contributing to data-driven financial analysis.

Dashboard Code

Height Calculation

A CCD camera was employed to capture in-situ images during the printing process. Images captured are processed using OpenCV to detect and extract the printed metal region via color masks in grayscale, RGB, and LAB color spaces. Height is estimated by counting pixels along user-defined vertical lines and converting them to physical measurements. The analysis helps monitor layer-wise growth and maintain the optimal nozzle-to-substrate distance. Python libraries like OpenCV, NumPy, and Matplotlib were used for image processing and visualization, and CSV files are generated for further analysis.

GitHub Link

BandGap Prediction

Predicted the band gap of perovskite materials through Ml model, providing a faster alternative to computationally expensive Density Functional Theory (DFT) calculations. Using data from the Materials Project and tools like PyMatGen, we built a dataset of stoichiometric perovskites (ABX₃) and extracted both compositional and structural descriptors. Band gaps were predicted using XgBoost and regression models (e.g., Extra Trees Regressor with R² ≈ 0.78). The project demonstrated how ML can support high-throughput screening of novel materials for electronic and energy applications.

GitHub Link

Self  Projects

Pizza Sales Analysis

Leveraged SQL for data extraction and transformation, and Power BI for dynamic visualization to analyze a full year (2015) of pizza sales data. I created interactive dashboards that highlighted key business insights, including total revenue (~$818K), average order value, and sales trends by day, month, pizza category, and size. The analysis identified that classic pizzas and large sizes were the most popular, with peak orders occurring on weekends. Additionally, ranked best and worst-selling pizzas by revenue, quantity, and order frequency, enabling strategic decisions for product optimization.

Dashboard