Welcome, I'm Luca Ostinelli

• 08/2026 - 02/2027 — Visiting Researcher, Saint Louis University, USA

• 05/2025 - 04/2026 — Master's Student Research Fellow – UROP Programme, Politecnico di Torino, Italy

  • Focused on the automatic generation of university course timetables through linear optimization techniques.

• 2024 - Present — M.Sc. in Computer Engineering, Politecnico di Torino, Italy
  • Thesis (in collaboration with Saint Louis University, USA)
• 2024 — B.Sc. in Computer Engineering, University of Pisa, Italy
  • Thesis (in Italian 🇮🇹)
• 2019 — High School Diploma in Computer Science

Check out my personal website for projects and my CV.

My publications can be found on this page.

• StaffScheduler

  Staff scheduling platform based on optimization algorithms

  Collaborative project aimed at optimizing staff scheduling. Developed “StaffScheduler”, a platform that permits managers to efficiently allocate shifts and track employee availability based on personal preferences.

• Bitcoin Academy

  Educational initiative on Bitcoin and blockchain technology

  BitPolito Academy is an open-source, local-first AI tutor that turns Bitcoin study materials into structured, interactive learning paths with lessons, quizzes, and progress tracking.

• Verilocal

  Decentralized crypto adoption platform (5-member team)

  Collaborative project aimed at promoting the use of cryptocurrencies among local businesses in Lugano, Switzerland. Developed “Verilocal”, a platform that connects crypto-friendly merchants with users, facilitating real-world adoption through verified listings and geo-localized discovery.

• Semantic Correspondence

  Semantic Correspondence (4-member team)

  We explore and compare multiple approaches for semantic correspondence estimation, leveraging foundational vision models such as DINOv2, DINOv3, and SAM. Implementations include dense feature extraction, spatial alignment via window soft-argmax, and efficient fine-tuning techniques using LoRA and selective layer unfreezing, analyzing trade-offs between matching accuracy, zero-shot generalization capabilities, and computational processing costs.

• 6D-Pose-Estimation

  6D Pose Estimation project (4-member team)

  We explore and compare multiple approaches for 6D object pose estimation, ranging from RGB-only methods (ResNet with pinhole camera model) to RGB-D fusion techniques. Implementations include MLP-based residual learning, DenseFusion with iterative refinement, keypoint voting (PVN3D), and bidirectional fusion (FFB6D), analyzing trade-offs between accuracy, computational cost, and modality integration strategies.

• Rust Project

  Cross-platform group chat application in Rust (4-member team)

  Built "Ruggine", a client-server chat system supporting user registration, group invitations, and message exchange. Focused on CPU usage efficiency and binary size optimization.

• Computer Architectures

  Pac-Man simulator in C and Assembly

  Developed a Pac-Man-style game simulator by combining low-level logic in C with custom assembler microprogramming targeting a simplified CPU architecture.

• Web Applications

  Full-stack SPA using Node.js, React, and SQLite

  Designed and implemented a task management platform for university use, featuring role-based access for students and instructors. Enforced complex group formation rules, weighted grading, and secure session-based API authentication via Passport.js.

• Software Engineering

  Distributed monitoring system for geophysical networks (4 members)

  Team project simulating real-world software engineering workflows. Developed "GeoControl", a layered system for sensor data collection and visualization. Backend written in TypeScript, secured with token-based API authentication, and documented via Swagger. Unit and integration tests implemented with Jest (by Meta).

• Human Computer Interaction

  Virtual Podium – Adaptive Feedback System for Public Speakers (3-member team)

  Developed "Virtual Podium", an application that assists public speakers by providing real-time feedback on posture and speech pace. Compatible with both smartphones and VR headsets. The development process started with user interviews to identify user needs, followed by paper prototyping and then digital prototyping to refine usability and interaction design.

• Computer Networks

  Socket programming in C

  Implemented a client-server communication system using TCP/IP sockets in C. Features robust protocol handling and message exchange logic. Includes a detailed project report explaining the architecture, protocol design, and testing methodology.