Marcos Sanson

Work Experience

• Selected for the competitive AFRL Scholars Program to join a research team focused on artificial intelligence for autonomous space systems
• Assigned to the project “Artificial intelligence spacecraft control for contested space applications using reinforcement learning”

• Collaborate with computer science professors on research in artificial intelligence (AI) and machine learning (ML)
• Develop and optimize evolutionary algorithms using experimental design, high-performance computing, statistical tools, and ML frameworks like PyTorch
• Co-authored a peer-reviewed paper published by ACM and presented at the Genetic and Evolutionary Computation Conference

• Led integration of new microcontrollers in light-sensing and auto-dimming vehicle mirrors to improve real-time processing
• Deployed embedded software, reviewed code, wrote design documentation, and conducted testing
• Worked with teams in software testing, project management, requirements engineering, and finance

• Provided technical support to over 3,000 university staff and faculty
• Installed and maintained computing equipment and resolved hardware/software issues
• Collaborated in team-based scheduling and resolution of multiple support requests

• Developed a Python-based meteorological application during my Intercultural Communication & Service Learning Spain Study Abroad program
• Automated generation of weather databases and environmental visualizations for real-world sustainability projects
• Implemented a multilingual tool (English, Spanish, Galician) with both a Command Line Interface (CLI) and Graphical User Interface (GUI), packaged as a standalone executable
• Improved data processing speed by over 200,000%
• Collaborated with an international team in-person in Spain and remotely from the United States

Projects

• 1. Identifying Evolvability-Enhancing Mutations via Computational Models October 2024 – Present (expected: May 2025)
  

  Research poster presented at GVSU Student Scholars Day

  • Overview: Traditional biological evolution experiments take months, years, or even decades to complete, but computational models allow us to study thousands of generations per minute. With this capability, I am researching how certain mutations improve evolvability using NK landscapes, a mathematical model of fitness landscapes.
  • Key Activities:
    • Designing computational experiments to analyze fitness data
    • Testing various evolvability metrics to quantify adaptability
    • Running large-scale simulations on Grand Valley State University's high-performance computing cluster
    • Comparing fitness landscapes with and without specific mutations to identify trends in evolvability
  • Impact: Identifying patterns that reveal how specific mutations enhance a population’s ability to adapt over time, contributing to a deeper understanding of evolvability in biological and artificial systems
  • Mentor: Dr. Austin Ferguson, Assistant Professor, College of Computing, Grand Valley State University
  • GitHub Repository: https://github.com/FergusonAJ/quantifying_evolvability
  • Acknowledgments: Supported by the Kindschi Undergraduate Research Fellowship (Winter 2025)
• 2. Automatic Atrium Segmentation via Machine Learning September 2024 – December 2024
  

  Research poster summarizing the final results of the atrial segmentation project. View full version on GitHub.

  • Overview: Traditional manual segmentation of atrial structures (regions of the heart) in medical imaging is time-consuming and varies between observers. Using machine learning, I developed a model for automatic 3D segmentation of the atrium to improve accuracy and efficiency in cardiac image analysis.
  • Key Activities:
    • Preprocessed 3D medical imaging data to ensure high-quality inputs
    • Built data pipelines to automate and streamline processing
    • Optimized mesh alignment using neural networks to improve segmentation accuracy
  • Impact: Improved the speed and precision of atrial segmentation, reducing manual effort and enhancing diagnosis and treatment planning in cardiology
  • Mentors:
    • Dr. Gonzalo Ricardo Ríos Muñoz, Assistant Professor, Bioengineering Department, Universidad Carlos III de Madrid (UC3M)
    • Dr. Pablo Martínez Olmos, Associate Professor, Signal Theory and Communications Department, UC3M
    • Dr. Antonio Artés Rodríguez, Professor, Signal Theory and Communications Department, UC3M
  • GitHub Repository: https://github.com/officialconfuzius/cardiologyml
• 3. Phylogenetic Analysis for Evolutionary Search Algorithms September 2023 – July 2024
  

  Poster presented at the 2024 Genetic and Evolutionary Computation Conference (GECCO)

  • Overview: I collaborated with Dr. Alexander Lalejini to enhance evolutionary search algorithms using phylogenetic analysis. We improved optimized problem-solving by integrating evolutionary biology concepts into computational methods.
  • Key Activities:
    • Designed and executed experiments on a high-performance computing cluster
    • Developed novel subsampling methods for optimization problems
    • Analyzed experimental data to assess algorithm performance improvements
    • Co-authored a peer-reviewed paper and presented our work at the 2024 Genetic and Evolutionary Computation Conference
  • Impact: Our work demonstrated that incorporating phylogenetic analysis into evolutionary search algorithms can significantly enhance their efficiency and effectiveness in solving complex optimization problems.
  • Mentor: Dr. Alexander Lalejini, Assistant Professor, College of Computing, Grand Valley State University
  • Publication: Runtime phylogenetic analysis enables extreme subsampling for test-based problems
  • GitHub Repository: https://github.com/amlalejini/phylogeny-informed-subsampling
  • Acknowledgments: Supported by the Kindschi Undergraduate Research Fellowship (Winter 2024)
• 4. Procedurally-generated 2D Role-playing Game January 2024 – April 2024
  

  Screenshot from final game build. View on GitHub.

  • Overview: In a software engineering course (CIS 350), I collaborated with a team to develop a 2D role-playing game featuring procedurally generated content. Our goal was to enhance replayability and player engagement by ensuring unique gameplay experiences with each session.
  • Key Activities:
    • Designed core game mechanics, including character progression and combat systems
    • Implemented procedural content generation algorithms to create dynamic environments and quests
    • Conducted playtesting sessions to gather feedback and refine gameplay elements
  • Impact: Delivered a functional game prototype that demonstrated the potential of procedural generation to enhance replayability in role-playing games.
  • GitHub Repository: https://github.com/Marcos-Sanson/Procedural-RPG
• 5. Meteorological Software Application May 2023 – November 2023
  

  Example GUI from the final application build. View full screenshot on GitHub.

  • Overview: During my study abroad work at Parque Náutico de Castrelo in Spain, I developed a Python-based application to automate the creation of weather databases and real-time data visualization. This project aimed to improve environmental monitoring and support projects like tree planting and erosion management.
  • Key Activities:
    • Developed data pipelines to automate the processing of multilingual datasets
    • Integrated data visualization tools to monitor environmental conditions
    • Collaborated with a cross-cultural team to ensure the application's effectiveness
  • Impact: Improved weather data processing speeds by over 200,000%, significantly increasing the efficiency of environmental and meteorological monitoring.
  • GitHub Repository: https://github.com/Marcos-Sanson/MeteoGalicia-Application
• 6. Integrated Machine Learning for Robot Control October 2022 – September 2023
  

  Dr. Jared Moore presenting on genetic algorithms and evolutionary robotics during the GVSU Computing Seminar Series

  • Overview: I collaborated with Dr. Jared Moore at Grand Valley State University to integrate machine learning and evolutionary algorithms for controlling a digital quadruped robot. Our goal was to enhance the robot's gait and control efficiency.
  • Key Activities:
    • Designed experiments and analyzed data to optimize algorithms
    • Pretrained neural networks using supervised learning before applying neuroevolution techniques
    • Refined hyperparameters to improve algorithmic performance
    • Developed tools for experiment execution and data analysis, ensuring reproducibility and integrity of results using frameworks like PyTorch
  • Impact: Improved the efficiency and adaptability of quadruped robot control systems, contributing to advancements in evolutionary robotics.
  • Mentor: Dr. Jared Moore, Associate Dean of Undergraduate Studies & Outreach, Grand Valley State University
• 7. Spot Micro Agile Robot Dog October 2021 – May 2022
  

  My project partner, Ronit Nagarapu, and I with the completed Spot Micro Robot Dog

  • Overview: Working with another student, I developed a fully operational robot inspired by Boston Dynamics' Spot agile mobile robot. Our goal was to create a cost-effective, functional quadruped robot for both disaster relief and research purposes.
  • Key Activities:
    • Developed Python-based software enabling remote control via Bluetooth
    • Integrated hardware components, including a Raspberry Pi, 3D-printed parts, servo motors, ultrasonic sensors, and an LCD display
    • Conducted extensive testing and debugging to ensure functionality across various environments
    • Published project documentation and contributed open-source code to the robotics community
  • Impact: Created a functional quadruped robot that serves as a platform for further research and development in robotics, and is still active to the present day.
  • Project Website: https://sites.google.com/view/senior-tech-project