Hi, I'm Leonor De Moura

LAPR4 – AISafe Flight Control & Simulation System

Enterprise-level software prototype that simulates large-scale air traffic operations, processes domain-specific flight plans and provides real-time collision detection and prevention.

Architecture & Enterprise Development

• Domain-Driven Design around core aviation concepts: aircraft, airports, routes and pilots.
• SCRUM across three sprints, with 90%+ unit test coverage.
• Flexible persistence layer supporting in-memory storage and remote RDBMS.

Concurrent Flight Simulation (C, Unix)

• Each flight runs as an independent process, communicating through POSIX pipes and coordinated with semaphores.
• Multi-threaded coordination layer with shared memory, POSIX threads, mutexes and condition variables.
• Real-time collision and hazard detection, notifying flights through Unix signals (SIGUSR1).

Custom DSL & Compiler

• Domain-Specific Language for describing and validating flight plans, with its grammar defined in ANTLR.
• Semantic validation through AST traversal (Listeners and Visitors) with structured compile-time error reporting.

Networking & Distributed Systems

• Secure TCP client-server architecture, keeping the database unexposed to clients.
• Cloud-based flight logging server using UDP datagrams for real-time telemetry.
• Web server with AJAX-based live status updates.

Images

LAPR2 – Railway Network Simulation (MVP)

Minimal Viable Product for the simplified simulation of railway operations, inspired by RailRoad Tycoon, with a Map/Scenario Editor and a Railway Simulator.

• Software development: SCRUM across three sprints, Java with strict TDD (JUnit 5) and a JavaFX 11 GUI.
• Network algorithms: connectivity, train feasibility based on electrification, maintenance routes and shortest routes through checkpoints, with worst-case complexity analysis.
• Statistics & KPIs: descriptive and comparative analysis (boxplots, histograms) and linear regression to predict revenue, in Python/Jupyter.

LAPR1 – Facial Recognition Using Eigenfaces

Java application (fictionally commissioned by the European Intelligence Agency) to represent and identify individuals using eigenfaces derived from facial images. Each grayscale image is treated as a matrix of values (0–255) and approximated with a subset of k eigenvectors.

• Eigen decomposition of symmetric matrices.
• Image reconstruction using the most relevant eigenfaces.
• Identification of the closest match in a database by comparing the Euclidean distance between weight vectors (Ω).