Innovative data enthusiast with a Certified Data Scientist credential from DataMites Global Institute and hands-on experience as a Data Science Consultant Intern at Rubixe. Proficient in Data Wrangling, Statistical Analysis, Machine Learning, Deep Learning, NLP, LLMs, Python, MySQL, PySpark, Git, AWS, and data visualization using tools like Tableau, and Power BI. Demonstrated ability to deliver actionable insights through multiple Proof of Concept (POC) projects and client assignments. Passionate about leveraging data-driven solutions to solve complex business problems and drive innovation.
AutoViz Dashboard is an advanced interactive data visualization app built with Streamlit and Plotly that transforms raw CSV, Excel, or JSON data into dynamic, professional dashboards. The application intelligently caches data to ensure rapid processing and performs automated data cleaning, allowing users to focus on extracting actionable insights. It offers a robust suite of charts-including histograms, line charts, bar charts, bubble charts, scatter plots, hexbin plots, treemaps, crosstab tables, count plots, box plots, pie charts, donut charts, time series, geospatial maps, and more-along with interactive filters and real-time Key Performance Indicators (KPIs). It's an ideal solution for data analysts & scientists, business leaders, researchers, and innovators seeking to convert complex data into clear, strategic visual narratives.
App Link
SmartML Pro is a Streamlit-based end-to-end machine learning app for data science professionals, supporting classification and regression tasks with data input via MySQL or file upload. It features comprehensive EDA, data preprocessing and a variety of ML models for both classification and regression, Hyperparameter tuning and evaluation metrics. The app integrates an AI assistant powered by Azure OpenAI GPT-4.1 accessible from the sidebar for help and troubleshooting. It enables rapid prototyping, automated data cleaning, handling imbalanced data with SMOTE, and interactive AI support to accelerate workflows and improve model performance in business use cases.
App Link
Developed a Retrieval-Augmented Generation (RAG) chatbot using Flowise AI for workflow orchestration, GPT-4o Mini as the generative language model, and Vectara for efficient vector storage and semantic search. Designed to process user-uploaded PDF files, the system extracts text, converts it into vector embeddings via Vectara, and retrieves contextually relevant data to generate accurate, context-aware responses using GPT-4o Mini. By prioritizing PDFs as the knowledge source, the chatbot enables structured document analysis for industries reliant on static documents (e.g., legal, academic). Ideal for knowledge management, research, or customer support, it transforms static data into dynamic conversational experiences.
GitHub Link
Developed a multilingual translator chatbot utilizing GPT-4o Mini and Flowise AI, capable of translating between English and various languages such as Tamil, Hindi, French, German and more. The chatbot allows users to customize translations for specific language pairs, enhancing its versatility. Designed to facilitate seamless communication across different languages, this user-friendly system integrates the advanced capabilities of GPT-4o Mini with Flowise AI's intuitive interface. This makes it an invaluable tool for global teams, travelers or anyone who require quick and efficient translations, ultimately breaking down language barriers and improving accessibility in communication.
GitHub Link
Developed a local AI chatbot using Flowise AI and the Ollama framework with the DeepSeek-R1 model, ensuring data privacy and customization for specific needs. This setup enables powerful conversational AI capabilities. Integrated advanced natural language understanding and generation through DeepSeek-R1, allowing the chatbot to handle complex queries and deliver accurate, context-aware responses. Flowise's user-friendly interface simplifies workflow design and management. Making it accessible for developers and enhancing overall user experience.
GitHub Link
Fabricated drive pulleys with artificial wear faults (pulley shaft wear and outer layer wear) and collected data under three conditions: good, pulley shaft wear, and outer layer wear using an accelerometer sensor. Extracted statistical features from the data and utilized MATLAB's Machine Learning and Deep Learning Toolbox to test Decision Tree, Artificial Neural Network, and Support Vector Machine (SVM) algorithms. The SVM algorithm achieved the highest accuracy of 94%, aiding in the prevention of unscheduled maintenance and breakdowns in industrial settings.
GitHub Link