Matteo Bertoia

This project includes two scripts:
hotmail_sent_email_extractor.py: Extracts sent emails from your Outlook/Hotmail account using Microsoft Graph to build a reply dataset (email_reply_dataset.json).
mac_graph_email_drafter.py: Uses a local LLM via Ollama to automatically generate Outlook email drafts based on incoming messages and your past replies.

Note: Both scripts require inserting your Microsoft Graph CLIENT_ID.

This project combines the intelligence of Large Language Models (LLM) with the flexibility of an AI Orchestrator and a modern React-based frontend to deliver an interactive and adaptive user experience.
LLM Layer: Powers natural language understanding and generation, enabling context-aware responses and reasoning.
AI Orchestration: Coordinates multiple AI services, tools, and retrieval pipelines (e.g., RAG) to provide accurate and domain-specific insights.
React Frontend: A dynamic… Visualizza altro

This project combines the intelligence of Large Language Models (LLM) with the flexibility of an AI Orchestrator and a modern React-based frontend to deliver an interactive and adaptive user experience.
LLM Layer: Powers natural language understanding and generation, enabling context-aware responses and reasoning.
AI Orchestration: Coordinates multiple AI services, tools, and retrieval pipelines (e.g., RAG) to provide accurate and domain-specific insights.
React Frontend: A dynamic, responsive, and modular interface that allows seamless user interaction with the AI system in real time.
The result is a scalable, modular, and intelligent platform capable of handling complex workflows, delivering human-like interactions while ensuring data grounding and explainability. Meno dettagli

Physics-Informed Neural Networks combine deep learning with physical laws.
Unlike traditional neural networks, which rely solely on data, PINNs incorporate known physical laws (usually in the form of PDEs) into their loss functions. This unique approach ensures physically consistent predictions, even when data is limited. PINNs adapt to the granularity of discretization and network structure, offering remarkable predictive accuracy.

It is a C# application to read and write Digital Input / Output from all the USB-5856 ADVANTECH devices, also automatically by follow some syntax rules needed to generate a list of commands. Also the User Manual is provided to understand all the functionalities

It is a C++ application that can be used to do automatic tests. First run your code X to be tested and read values 1 => ON, 0 => OFF for some buttons depending on your machine (by default there are 4 buttons implemented), then modify their values locally both automatically or manually through send_udp.exe! In automatic mode it is also possible to control the timing between each send. You can find an example in file command.txt. For example "1 0 0 0;1000" mean that you want BUTTON 1 =>… Visualizza altro

It is a C++ application that can be used to do automatic tests. First run your code X to be tested and read values 1 => ON, 0 => OFF for some buttons depending on your machine (by default there are 4 buttons implemented), then modify their values locally both automatically or manually through send_udp.exe! In automatic mode it is also possible to control the timing between each send. You can find an example in file command.txt. For example "1 0 0 0;1000" mean that you want BUTTON 1 => ON, BUTTON 2 => OFF, BUTTON 3 => OFF, BUTTON 4 => OFF and you will wait 1000 ms before sending the next values. Obviously, the ports for sending packages must to be configured based on your needs; by default are made two sends for ports 3001 and 4001. The receive_udp.exe is useful to test first the UDP communication; it must be implemented directly inside the code you need to test. It is ready to be customized based on your needs Meno dettagli