Work Packages


WORK PACKAGES

Project Structure

The project is organized into 5 work packages: a shared platform, three vertical applications addressing the partners’ use cases, and a work package dedicated to dissemination.

01
WP1
LEAD ORGANIZATION: UNICAL

SIGENERA Platform

Critical analysis of GenAI/LLM techniques and implementation of the shared platform.

A1.1

State-of-the-Art Analysis

omprehensive analysis of available GenAI/LLM technologies: fine-tuning, RAG, quantization, testing techniques, and methods for correcting undesirable behaviors.

A1.2

Platform Implementation

Design and implementation of the SIGENERA platform: scalable, featuring fine-tuning, querying, and advanced prompting techniques on GPUs.

02
A2.1

LLM Models for the Help Desk

Study of the theoretical foundations of Large Language Models (LLMs) for automated response systems. Customisation of the SIGENERA platform using data from the University of Calabria's Student Services, serving up to 50,000 students.

A2.2

UNICAL Chatbot Development

Development of the chatbot to handle student enquiries, resolve common issues automatically, and route more complex requests to human operators, following an Agile development methodology.

WP2
LEAD ORGANIZATION: UNICAL

University Help Desk

Adaptation of the SIGENERA platform to the Help Desk chatbot for UNICAL’s Educational Services M7–M24 · Type: RI, SS

03
WP3
LEAD ORGANIZATION: Digiservice

Source Code Generation

Adaptation of the platform for querying and monitoring information systems, applied to TimeSheet Connect and MyVisure. M7–M24 · Type: RI, SS

A3.1

LLM Models for Information Systems Querying

Adaptation of LLM models for generating queries and commands in the DML of the MYVISURE information system (Tecnoinnovis) and TimeSheet Connect Q (Digiservice).

A3.2

Software Services Development and Validation

Development of software modules to integrate LLM models into the target platforms. ITS CADMO interns will be involved in on-the-job training. AGILE methodology.

04

A4.1

LLM Models for Meeting Minutes Generation

Adaptation of LLM models for the automatic generation of meeting minutes within the EligoVote platform (ID Technology), using AI techniques to improve the accuracy, consistency, and reliability of the content.

A4.2

Software Services Development and Validation

Development of software services to integrate LLM models into the EligoVote platform through SiGENERA. The prototype will be validated using a dataset of transcripts and voting records. AGILE methodology.

WP4
LEAD ORGANIZATION: ID Technology

Meeting Minutes Generation

Verticalizzazione della piattaforma SIGENERA per la generazione automatica di verbali di riunione nella piattaforma EligoVote, mediante modelli LLM specializzati, tecniche di AI per il supporto contestuale e servizi software integrati per la produzione e la validazione dei verbali. M7–M24 · Type: RI, SS

05
WP5
LEAD ORGANIZATION: TECNOINNOVIS

Dissemination of Results

Sharing the project’s results with the scientific community and the general public through multiple channels. M1–M24 · Type: RI, SS

A5.1

Communication and Project Website

Regional workshop, brochure, infographics, bilingual website (IT/EN), and promotion through professional social media channels, particularly LinkedIn.

A5.2

International Conferences

Participation in at least two international conferences in different EU Member States. Publication of articles in specialized ICT and FinTech journals

Five work packages,
one shared vision.
Bringing Artificial Intelligence to serve people, organizations, and society.


Discover the Project
→