Collection of projects developed at Emflora Regenerate using:

Responsible for initiating the implementation and application of Data Science strategies to enhance company results.

Some of the projects developed:

- Time Card Information Extraction Automation via OCR
Developed an OCR pipeline to automatically extract and process information from time cards and integrate it into the company's system;
Reduced processing time and minimized manual errors in employee time management;

- Intelligent Chatbot for Internal Support
Modeled an intelligent chatbot trained on the company manual, utilizing LLMs with optimized prompts, Transformer-based architecture via Hugging Face, RAG (Retrieval-Augmented Generation) approach, Redis for context caching, and Gunicorn as the application server for production;

- Inventory Prediction and Purchasing Optimization
Built predictive models using Prophet, XGBoost, LSTM, RNN, and PyTorch to forecast inventory levels and identify the optimal reorder point;
Predicted future expenditures based on historical consumption, reducing costs from unnecessary purchases and preventing stockouts;
Visualized results in Power BI to facilitate decision-making.

- Financial Prediction by Department
Built a predictive model to estimate revenue and costs per department for the following month;
Utilized XGBoost, Prophet, LSTM, RNN, and PyTorch to model different financial behavior patterns;
Implemented a data pipeline with PostgreSQL and ELT processes, ensuring efficiency in processing and integration with Power BI.

- Comprehensive Power BI Dashboards for Analysis