Alessio Saladino
EDUCATION
Bachelor's Degree
Computer Science at Università degli Studi di Bari "Aldo Moro"
Final Grade: 110L
Period: September 2017 – December 2020
Thesis Title: Social Robot and e-Health: Aphel as an Assistant for Patients in Hospital Wards
Thesis Description:
Development of an intelligent system that allows hospital patients to communicate and request support from the social robot Aphel through IoT technologies. The robot is capable of understanding the patient's needs and providing information or assistance. If necessary, the robot can remotely alert medical staff for direct intervention.
Master's Degree
Artificial Intelligence and Robotics at Sapienza University of Rome
Final Grade: 104
Period: September 2021 – October 2024
Thesis Title: A Novel Architecture Integrating LLM and Context-Aware Supervisors for Safe Interaction with Social Robots
Thesis Description:
Development of a cognitive architecture for human-robot interaction. The architecture enhances the robot's cognitive capabilities by leveraging information from long-term memory and the use of Supervisors that manage interaction flow. The context retrieved from long-term memory via Retrieval Augmented Generation (RAG) is processed by LLM-based supervisors, allowing the robot to act safely and appropriately in context.
WORKING EXPERIENCES
Salesforce Developer Consultant at Sidea Group S.r.l.
Period: April 2021 – September 2021
Job Description:
Design, development, and maintenance of customized solutions to support business processes.
Technical skills acquired during this period include:
- Development of custom solutions using Apex programming language, including classes, triggers, batch processes, and scheduled jobs, with a focus on automating complex business workflows;
- Design and management of integration between Salesforce and external systems via REST and SOAP API calls, using Postman for client-server communication testing and validation;
- Use of version control tools (Git) to ensure traceability of changes and manage releases across different environments (sandbox and production);
- Full development lifecycle involvement: analysis, implementation, test classes, deployment, and post-release maintenance;
- Use of Salesforce-integrated DBMS for enterprise data management and querying;
- Development with Visual Studio Code integrated with Salesforce for local development and controlled deployment;
- Team collaboration following Agile methodologies.
DISSEMINATION
During my PhD studies, I took part in outreach events aimed at a diverse audience. Below are the events I participated in:
- OpenDIAG 2025: Outreach event focused on guiding high school students in their university choice. I contributed by presenting a hands-on demo involving the social robot MARRtina, enhanced with the Large Language Model Minerva to demonstrate context-aware, natural human-robot interaction.
- GardAI 2025: Public engagement event centered on the role of Artificial Intelligence and Robotics in socially impactful scenarios. I presented a demo where the NAO and SMARRtino robots engaged in collaborative dialogue, leveraging Large Language Models to simulate autonomous inter-robot communication.
ACADEMIC PROJECTS
During my studies, I had the opportunity to work on various projects related to Human-Robot Interaction, industrial robotics, and the training of artificial intelligence models using the PyTorch framework for Python.
In addition, I gained experience in scientific writing and academic research.
Title: Intestinal Metaplasia Detection
Course: AI for Visual Perception in HCI and HRI
Period: February – May 2024
Description:
Patch-based classification of intestinal metaplasia from medical images. The approach includes a voting mechanism across extracted patches and a custom loss function to enable pseudo-segmentation.
Technologies: PyTorch
Skills: Histogram analysis, Image classification, Scientific writing, Pseudo-segmentation
Title: Split Computing and Knowledge Distillation
Course: Neural Networks
Period: December 2023 – February 2024
Description:
Implementation of teacher-student architectures to optimize the computational performance of neural networks using knowledge distillation and split computing techniques.
Technologies: PyTorch
Skills: Knowledge Distillation, Split Computing, Report Writing
Title: Question Answering on Mathematical Datasets
Course: Deep Learning
Period: August – September 2023
Description:
Custom implementation of a Transformer architecture for question answering on a mathematical dataset, featuring a tensor-product attention mechanism.
Technologies: PyTorch
Skills: NLP, Generative AI, Transformer Architecture, Teamwork
Title: Pepper Robot as a Road Safety Teacher
Course: HRI + Reasoning Agents
Period: June – November 2023
Description:
Design of an interactive educational application for the Pepper social robot, integrating lessons and games on road safety, with a child-friendly interface.
Technologies: PyTorch, PDDL, Android SDK
Skills: UX, Robot Programming, Child-friendly UI
Title: Survey Papers – NLP & AI Ethics
Courses: NLP, AI & Robotics Seminars
Period: May – July 2023
Description:
Writing of survey papers on text summarization techniques, explainability, and ethical challenges in AI.
Technologies: LaTeX, Overleaf
Skills: Academic writing, Scientific paper review, AI ethics
Title: AI Quest Generator for Role-Playing Games
Course: Narrative Understanding & Storytelling
Period: May – June 2023
Description:
Fine-tuning of a Large Language Model (GPT-2) to generate role-playing game quests. The dataset was built through web scraping, and outputs were evaluated via qualitative analysis.
Technologies: PyTorch, Regex, GPT-2
Skills: Generative AI, Dataset building, Web scraping
Title: 3D Robotic Manipulator in Virtual Environment
Course: Robot Programming
Period: August – September 2022
Description:
Robotic simulation in Gazebo with both forward and inverse kinematics control.
Technologies: C++, ROS, Gazebo
Skills: Robotic simulation, Manipulation, ROS
Currently enrolled in the 40th cycle of the National PhD in Artificial Intelligence at Sapienza University of Rome, under the supervision of Prof. Luca Iocchi and in international collaboration with Procter \& Gamble. My current research focuses on enhancing the cognitive capabilities of robotic systems by leveraging the power of Large Language Models (LLMs). The impressive language processing abilities of modern LLMs make them valuable tools in social robotics, where they can significantly improve interaction quality by enabling robots to understand their surrounding context and make safe, coherent, and socially appropriate decisions. These cognitive functionalities can also be extended to industrial domains, allowing the integration of a cognitive architecture aimed at increasing the efficiency and productivity of industrial processes. By exploiting cognitive capabilities, a robot can adapt its behavior based on the context in which it operates. This enables the application of the same cognitive architecture across multiple domains, with only the contextual descriptions and information needing to be adjusted.