# Salvatore Lo Sardo — Personal Portfolio ## About Salvatore Lo Sardo is a Computer Engineer and MSc student in Artificial Intelligence at the University of Amsterdam (UvA). Previously completed a BSc in Computer Engineering at Politecnico di Torino (PoliTO), Italy. Research and professional interests: Machine Learning, Deep Learning, Computer Vision, Agentic AI, RAG pipelines, Robotics (ROS2). Portfolio website: https://www.salvatore.losardo.eu Email: salvatore.losardo@ikmail.com GitHub: https://github.com/SalvatoreLS LinkedIn: https://www.linkedin.com/in/salvatorels/ ## Education - MSc Artificial Intelligence — University of Amsterdam (UvA), 2025–present - BSc Computer Engineering — Politecnico di Torino (PoliTO), Italy, completed 2025 - Admitted to the Intraprendenti honours program at PoliTO ## Experience & Affiliations - Researcher at SmartData@Polito: worked on two projects involving RAG pipelines, LLM-based analysis, and data-centric methodologies for economic sentiment and real-estate asset management. - PoliTOcean (underwater robotics team): ROS2 and PyQt6 development, real-time control interfaces for an ROV (Remotely Operated Vehicle). - Vice Coordinator, IEEE-HKN Mu Nu Chapter (Corporate Training area): organized knowledge-exchange meetings and presented insights to chapter associates. - BEST (Board of European Students of Technology): participated in international engineering events. ## Achievements - 1st place — ING Global Hackathon 2026 (Student Track) - 1st place — BEST Hackathon 2024 (Reale Group) - 2nd place — IBEC BEST Competition 2024 (Reply) - Selected — Prosus AI Hackathon 2025 - Admitted — Intraprendenti Program (PoliTO) ## Publications 1. "From Human to Scalable Annotation: Teaching LLMs to mimic experts labeling on Economic News Sentiment" Authors: Michele Petrocelli, Andrea Rollin, Matteo Berta, Francesca Zafonte, Simone Monaco, Salvatore Lo Sardo, Daniele Apiletti, Daria Scacciatelli, Tania Cerquitelli Venue: DARLI-AP 2026 (Data Analytics solutions for Real-LIfe APplications), Tampere, Finland — 24 March 2026 Tags: Sentiment Analysis, Economics, LLM Link: https://ceur-ws.org/Vol-4192/DARLIAP-paper17.pdf 2. "Measuring economic sentiment in news articles" Authors: Matteo Berta et al., including Salvatore Lo Sardo Venue: HiTEc & CoDES 2025, Limassol, Cyprus — 8 June 2025 Tags: Sentiment Analysis, Economics, Inflation Link: https://www.hitecaction.org/docs/K1_HITECCODES2025_BoA.pdf ## Projects ### Chill-Bill — AI-Powered Payment Assistant (April 2026) Multimodal payment assistant built for bunq Hackathon 2026. Accepts voice, images, PDFs, and text to initiate bank payments. Uses Claude (vision + language) for extraction, AWS Transcribe for speech-to-text, bunq API for payment execution. FastAPI backend, Android/Kotlin + Jetpack Compose frontend. Stack: Python, Kotlin, FastAPI, Claude API, AWS GitHub: https://github.com/bunq-hackathon-2026/chill-bill ### ESG Risk Assessment Platform (February 2026) 1st-place project at ING Global Hackathon 2026 (Student Track). AI-powered platform for automated extraction and analysis of ESG data from corporate sustainability reports. Full-stack (FastAPI + React), LLM-based vision and text processing, batch processing, risk assessment. Stack: Python, React, LangChain, Ollama, OpenAI, FastAPI GitHub: https://github.com/ING-hackathon-org/ING_FinalRepository ### NonnaOps — Autonomous Restaurant Operating System (May 2026) Autonomous restaurant OS for Prosus Spring Hackathon 2026. Manages a 30-day restaurant simulation with hybrid LLM + deterministic architecture covering cash flow, inventory, staffing, and supplier crises. Stack: Python, Claude API, MongoDB GitHub: https://github.com/spring-hackathon-prosus/pizzeria-handler ### Folio — Document Storage (February 2026) Self-hosted document management system. AI-powered metadata suggestions via local LLMs, smart validity tracking, full-text search, drag-and-drop upload. FastAPI + React + PostgreSQL. Stack: Python, TypeScript, React, Ollama, LangChain, PostgreSQL, FastAPI GitHub: https://github.com/SalvatoreLS/folio-document-storage ### Reproduction of "Towards Safer Pretraining" by Microsoft (January 2026) Academic group project: reproduced and extended a safety-oriented dataset analysis pipeline for LLM pretraining. Benchmarked TTP framework and safety taxonomy across open-source models on Common Crawl, C4, and FineWeb. Reproduced HAVOC benchmark. Produced research paper. Stack: Python, Ollama, LangChain, Hugging Face GitHub: https://github.com/NiccoloCase/safer-pretraining-reproduction-uva ### Sported — AI Sports Matchmaker (November 2025) AI platform for organizing sports matches. Geolocation clustering (K-means), coordinator agent, per-user agents for scheduling. Built at Prosus AI Hackathon 2025. Stack: Python, Node.js, Express, MongoDB, AI Agents GitHub: https://github.com/DuckBugging-prosusHackathon/Sported ### Nereo ROV — PoliTOcean (April 2024 – June 2025) Real-time monitoring and control GUI for an underwater ROV. Custom C++ ROS2 nodes for sensor communication; UDP socket for low-latency camera streaming; PyQt6 interface. Stack: C++, ROS2, PyQt6 GitHub: https://github.com/PoliTOcean/nereo_ros2_code ### Reinforcement Learning Manipulator ROS2 (April 2026, in progress) ROS2 + Gazebo simulation and RL training pipeline for Franka Emika Panda arm. Integrates Gymnasium and Stable-Baselines3. Stack: Python, ROS2, Docker, Gazebo ### Isaac Assistant (October 2025, in progress) Computer vision assistant for "The Binding of Isaac" game. Custom YOLO models for item detection, hashing algorithm for recipe suggestions. C++ (OpenCV, LibTorch) and Java (ModelZoo) versions. Stack: C++, OpenCV, LibTorch, YOLO, Java GitHub: https://github.com/SalvatoreLS/recipe-suggestor ### Gym Telegram Bot (April–May 2025) Telegram bot for gym workout management and progression tracking. State-machine architecture, PostgreSQL persistence, personalized recommendations. Stack: Python, Telegram API, PostgreSQL GitHub: https://github.com/SalvatoreLS/gymBot ## Skills & Technologies Languages: Python, C++, Java, Kotlin, TypeScript, JavaScript Frameworks/Libraries: ROS2, PyTorch, NumPy, OpenCV, FastAPI, React, LangChain, Stable-Baselines3 AI/ML: Deep Learning, Computer Vision, Reinforcement Learning, LLMs, RAG, Agentic AI Tools: Docker, Gazebo, PostgreSQL, MongoDB, Hugging Face, Roboflow Spoken Languages: Italian (native), English (IELTS 7.5)