{"componentChunkName":"component---src-pages-archive-js","path":"/archive/","result":{"data":{"allMarkdownRemark":{"edges":[{"node":{"frontmatter":{"date":"2025-08-13","title":"Farmona - Smart Crop Prediction & Agronomic Analysis","tech":["Python","Scikit-learn","NumPy","Pandas","Matplotlib","Streamlit"],"github":"https://github.com/Suborno-Deb-Bappon/Farmona","external":"","company":"Self-Initiated"},"html":"<p>Built an end-to-end crop classification system delivering 99.32% accuracy (weighted F1 = 0.9932) on a 440-sample holdout across 22 crops—just 3 misclassifications.</p>\n<ul>\n<li>Engineered 5 agronomic indices (THI, NBR, WAI, PP, SFI) and ran GridSearchCV over 7 algorithms with ≈420 hyperparameter fits to select the best leakage-safe Pipeline.</li>\n<li>Streamlit app supports one-click inference with top-5 class probabilities and schema-driven default auto-fill; artifacts packaged as 3 files (model, label encoder, schema) for reproducible rollout. </li>\n<li>Produced full classification report and plots (confusion matrix, permutation importance, learning curves) to verify generalization and pinpoint edge cases (e.g., rice recall 0.95 in holdout).</li>\n</ul>"}},{"node":{"frontmatter":{"date":"2025-08-05","title":"Blinkit Sales Analysis Dashboard","tech":["Power BI","DAX","Power Query","Excel"],"github":"https://github.com/Suborno-Deb-Bappon/Blinkit-Sales-Analysis-Dashboard","external":"","company":"Self-Initiated"},"html":"<p>Built a dynamic Power BI dashboard analyzing sales data from Blinkit’s retail outlets.</p>\n<ul>\n<li>Modeled and visualized multi-dimensional data using Power BI, DAX, and Power Query to enable real-time filtering by outlet type, sales volume, and customer ratings.  </li>\n<li>Identified key trends like low-fat product preference and the high profitability of Tier 3, medium-sized outlets.  </li>\n<li>Delivered actionable insights through calculated measures and interactive visuals to support data-driven retail strategy.</li>\n</ul>"}},{"node":{"frontmatter":{"date":"2025-07-20","title":"Crashlytics – Traffic Accident Severity Prediction","tech":["Python","Pandas","NumPy","Scikit-learn","Matplotlib","Seaborn"],"github":"https://github.com/Suborno-Deb-Bappon/Crashlytics","external":"","company":"Self-Initiated"},"html":"<p>Built a scalable machine learning pipeline to predict traffic accident severity using 7M+ US traffic records.</p>\n<ul>\n<li>Engineered features and applied boosting algorithms to achieve 85%+ prediction accuracy.  </li>\n<li>Analyzed spatio-temporal and environmental patterns to identify high-risk zones and peak periods, enabling 10–15% potential accident reduction.  </li>\n<li>Developed a reusable data science framework for real-time predictions and dashboards, supporting proactive traffic safety initiatives.</li>\n</ul>"}},{"node":{"frontmatter":{"date":"2025-05-10","title":"DEBug.me — Personal RAG-Powered Agentic AI Assistant","tech":["Python","Gemini API","OpenAI Embeddings","LangChain","Chroma (RAG)","NTFY","Gradio"],"github":"https://huggingface.co/spaces/suborno/DEBug.me","external":"","company":"Self-Initiated"},"html":"<p>Built and deployed DEBug.me, a personal agentic AI assistant that answers questions about my career using a vectorized knowledge base and multi-API integration.</p>\n<ul>\n<li>Implemented a Retrieval-Augmented Generation (RAG) system combining Chroma vector search with OpenAI and Google Gemini 2.5 Flash to deliver accurate, context-grounded answers from LinkedIn, resume, and summary documents.</li>\n<li>Implemented autonomous tool handling for user detail capture and unknown question logging, with real-time push notifications via NTFY for follow-ups.</li>\n<li>Deployed a Gradio-powered web app on Hugging Face Spaces, supporting real-time chat, conversation history, and continuous knowledge base enrichment.</li>\n</ul>"}},{"node":{"frontmatter":{"date":"2024-07-05","title":"Coffee Shop Sales Dashboard","tech":["Excel","Power Query","PivotTables","Charts"],"github":"https://github.com/Suborno-Deb-Bappon/Coffee-Shop-Sales-Dashboard","external":"","company":"Self-Initiated"},"html":"<p>Created an Excel-based dashboard analyzing $698K+ sales data for a retail coffee chain.</p>\n<ul>\n<li>Visualized KPIs (Avg Bill/Person, Orders/Hour, Footfall) to track trends across time, product, and store.</li>\n<li>Used slicers, calculated fields, and charts for interactive exploration.</li>\n<li>Delivered clear, data-backed insights on top products and peak business hours.</li>\n</ul>"}},{"node":{"frontmatter":{"date":"2024-07-05","title":"Job Market Analytics Using SQL","tech":["PostgreSQL","SQL","Excel","Data Aggregation"],"github":"https://github.com/Suborno-Deb-Bappon/Data-Job-Market-Insights","external":"","company":"Self-Initiated"},"html":"<p>Designed and implemented an SQL-based data analysis workflow to extract labor market insights across salary trends, demand, and key skills.</p>\n<ul>\n<li>Used advanced SQL queries including CTEs, joins, and aggregations to analyze multi-dimensional job data.</li>\n<li>Produced reusable, well-documented query scripts that reinforce best practices in integrity and performance.</li>\n<li>Translated findings into clear insights through GitHub documentation for both technical and non-technical users.</li>\n</ul>"}}]}},"pageContext":{}}}