PROJECTS

Still in Progress….🏗

SQL Genius

• Description: Developed an AI-powered SQL query agent that translates natural language into SQL, enabling seamless database interactions for non-technical users. The application allows users to upload SQLite databases, ask questions in plain language, and receive accurate SQL queries and results, simplifying data analysis and decision-making.
• Technologies (Libraries) Used: Python (Streamlit, Pandas), OpenAI API, SQL (PostgreSQL, SQLite)
• Generative AI Integration: Integrated OpenAI’s language models to dynamically interpret natural language queries and generate SQL, enhancing accessibility and usability.
• Live App: SQL Genius
• GitHub: GitHub Repository

🦅 Airfield Hazards: Bird Tracking at Airports

• Description: Developed a robust, cost-effective object detection pipeline for real-time bird identification at airports, using models like RetinaNet and YOLOv8, with advanced preprocessing techniques to enhance performance and safety.
• Reference Paper: AirBirds: A Large-scale Challenging Dataset for Bird Strike Prevention in Real-world Airports
• Technologies (Libraries) Used: Python (Altair, Matplotlib, PyTorch, Pandas, NumPy, Pytest), Makefile, AWS
• Models Tried: RetinaNet, YOLOv8, FasterR-CNN
• Project Report: Best Talk in UBC MDS
• Project Pipeline

• Interaction between the Scripts

Churn Insights

• Description: A project that involved key steps such as performing the ETL process and data cleaning in PostgreSQL, followed by data transformations and creating enhanced visualizations in Tableau. We then built and evaluated various machine learning models, including KNN, Decision Tree, Random Forest, RBF SVM, and Logistic Regression, in Jupyter Notebook. After conducting thorough EDA, we selected the best model to make predictions and present the results.
• Technologies (Libraries) Used: Python (Altair, Matplotlib, Plotly, PyTorch, Seaborn, Pandas, NumPy), Tableau, PostgreSQL
• GitHub: GitHub Repository

HomeScope

• Description: An analytical platform dedicated to illuminating the real estate market’s complexities. Aimed at stakeholders such as investors, developers, market analysts, and urban planners, it provides actionable insights through the careful analysis of pivotal variables influencing property values.
• Technologies (Libraries) Used: Python (Altair, Plotly, Pandas, PyArrow), Dash
• GitHub: GitHub Repository
• Link: Dashboard Website

CryptoPulse

• Description: A Shiny dashboard that allows users to interactively explore cryptocurrency data. It provides insights into various cryptocurrencies, with a focus on Bitcoin and Ethereum.
• Technologies (Libraries) Used: Python (Plotly), R (dplyr), Shiny
• GitHub: GitHub Repository

Pyxplor

• Description: A comprehensive Python package designed to automate and streamline the Exploratory Data Analysis (EDA) process. Tailored for various data types including numeric, categorical, binary, and time series data, pyxplor aims to enhance data interpretation through a suite of specialized plotting functions. This package seeks to reduce the complexity and time invested in initial data analysis, making it an essential tool for data scientists and analysts at all levels.
• Technologies (Libraries) Used: Python (PyPI, Pytest, Seaborn, Pandas), Poetry, Cookiecutter
• GitHub: GitHub Repository
• Tutorial: Docs