A conversational machine learning Discord bot that transforms various inputs into interactive models, visualizations, and data-driven insights.

Transformi positions itself as a data science proof-of-concept: a Discord-based analytic co-pilot that blends model experimentation, visualization, and deployment-ready application design.

---

The Vision

Modern analytics often live in fragmented tools: notebooks, BI dashboards, and command-line scripts. Transformi brings this experience into a single, conversational interface. The project demonstrates how a data scientist can interact with a dataset through chat: generating regression plots, selecting features, training models, and receiving visual artifacts without leaving Discord.

Engineering Highlights

• Conversational ML integration: Discord commands and rich UI components enable dynamic dataset selection, manual input, and on-demand model execution.
• End-to-end modeling: Supports linear regression, classification, neural network training, and ensemble methods using scikit-learn and tensorflow.
• Automated visualization: Generates charts and model outputs as image attachments, allowing users to interpret results immediately.
• Deployment-friendly health layer: A lightweight Flask service provides /, /health, and /ping endpoints to support managed hosting and uptime monitoring.
• Context-aware interaction state: User session locking and cached state management help the bot coordinate multi-step analytical workflows safely.
• Robust data handling: Uses pandas, numpy, and pyarrow for rapid preprocessing, dataset transformation, and CSV support.

Technical Architecture

• pandas, numpy, pyarrow for data ingestion and transformation
• scikit-learn for regression, classification, preprocessing, and evaluation
• tensorflow-cpu for neural network training and model visualization
• matplotlib, seaborn for charts and analysis plots

What Transformi Demonstrates

• Rapid exploration of raw and structured data through chat-driven inputs
• Visual regression analysis from generated or user-provided values
• Neural network training with immediate feedback and rendered artifacts
• Hybrid application design that combines bot UX with service monitoring

---