Federico Semeraro

Hello and welcome to my portfolio!

I am a Machine Learning Engineer within Apple's Video Computer Vision group, specializing in graphics and 3D generative AI.

Prior to this role, I worked as a Research Engineer at NASA Ames for nearly seven years, where I led the development of PuMA, the 2022 NASA Software of the Year. R&D efforts primarily focused on the segmentation and analysis of CT data to predict material properties for heatshields.

My past experiences include internships at various companies such as Airbus, Earth-i, and Planet Labs, which shaped a strong passion for computer vision. Projects ranged from analyzing aircraft movements, to extracting insights from satellite imagery, to natural scene generation. I am well-versed in several coding languages, especially Python, and currently extending my knowledge of PyTorch while implementing the latest deep learning techniques.

Academically, I completed a Master in Computer Science with a specialization in Machine Learning from the Georgia Institute of Technology in August 2023. I graduated from Imperial College London in June 2018 with a Master in Aerospace Engineering.

In my free time, you can find me on a tennis court, soccer field, or hiking somewhere around the Bay Area.

Whether it's to discuss new ideas, opportunities, or even just chat over a cup of coffee, please don't hesitate to reach out!

Skills

Core Competencies

Machine Learning Research; Computer Vision; Graphics and Neural Rendering; 3D Reconstruction; High-Performance Computing; Software Engineering

Tech Stack

PyTorch, USD, Blender, CUDA, Metal, Qt, Docker, AWS, CI/CD, Git, Conda/Pip

Coding Languages

Python (Proficient), Swift, C/C++ (Working Knowledge)

Languages

Italian (Native), English, Spanish (Fluent)

CV

Work Experience

Machine Learning Engineer

Apple

Oct 2024 – Present

Working in the Algorithms team of the Video Computer Vision group, specializing in graphics and 3D generative AI

Research Engineer

NASA Ames Research Center

Jan 2018 – Oct 2024

Led the development of ML tools and computational physics methods for the analysis of heatshield materials via CT data

Computer Graphics Intern

Planet Labs

Jun 2022 – Sep 2022

Automated the procedural generation of 3D synthetic Blender scenes to evaluate camera rendering models

Data Analytics Intern

Earth-i

Jul 2017 – Sep 2017

Trained deep learning models to analyze aerial imagery and predict hard-commodity production rates

Software Engineering Intern

Rolls-Royce

Jul 2016 – Sep 2016

Designed and deployed an internal database to track critical modifications to aircraft engine components

Year in Industry in Data Science

Airbus

Jul 2015 – Jun 2016

Implemented AI techniques to predict maintenance events using large-scale aircraft movement data

Education

Master of Computer Science

Georgia Institute of Technology

Jan 2020 – Aug 2023

Specialization in Machine Learning | GPA: 3.9

Master of Aerospace Engineering

Imperial College London

Sep 2013 – Jun 2018

Specialization in Computational Physics | GPA: 4.0

Awards

Innovation Award

NASA Ames Research Center

Dec 2023

For providing critical CT analyses supporting Artemis lunar mission and Mars Sample Return

Software of the Year

NASA

Jul 2022

Awarded to PuMA, adopted by over 50 research institutes and 150+ users

Article

Project Merit Prize

Imperial College London

Jun 2018

Best master thesis of the Aerospace Engineering cohort

Certificate of Excellence

Imperial College London

Jun 2015

Top performance at Imperial College London

Projects

Research

Apple MLR 2025

DSplats: 3D Generation by Denoising Splats-Based Multiview Diffusion Models

Authors: Kevin Miao, Harsh Agrawal, Qihang Zhang, Federico Semeraro, Marco Cavallo, Jiatao Gu, Alexander Toshev

A novel approach for generating 3D objects using splat-based multiview diffusion models, enabling high-quality 3D content creation.

Publication

Advanced Materials 2025

Carbon Fiber Oxidation in 4D

Authors: Benjamin M. Ringel, Federico Semeraro, Joseph C. Ferguson, Harold S. Barnard, Bruno Dias, Christian M. Schlepütz, Edward S. Barnard, Sam Schickler, Kara Levy, Shawn Shacterman, Talia Benioff-White, Julian Davis, Alastair A. MacDowell, Dilworth Y. Parkinson, Francesco Panerai

Analysis and simulation of time-resolved X-ray microtomography revealing oxidation-induced degradation in carbon fiber heat shields, impacting material properties and aerospace design.

Publication

CMAME 2025

Enabling FEM-based absolute permeability estimation in giga-voxel porous media with a single GPU

Authors: Pedro Cortez Fetter Lopes, Federico Semeraro, André Maués Brabo Pereira, Ricardo Leiderman

Open-source lightweight GPU solvers for FEM simulation of Stokes flow in porous media.

Publication

Spacecraft and Rockets 2025

arcjetCV: an open-source software to analyze material ablation

Authors: Alexandre Quintart, Magnus Haw, Federico Semeraro

A python package to automate time-resolved measurements of heatshield material recession from arcjet videos via edge detection and tracking.

Publication Github

SoftwareX 2025

TomoSAM: Slicer's extension for 3D segmentation

Authors: Federico Semeraro, Alexandre Quintart, Sergio Fraile Izquierdo, Joseph C. Ferguson

An extension of Slicer using the Segment Anything Model (SAM) to aid the segmentation of 3D data from tomography or other imaging techniques.

Publication Github Video Tutorial

Computational Materials Science 2023

Simulation toolkit for digital material characterization

Authors: Pedro C.F. Lopes, Rafael S. Vianna, Victor W. Sapucaia, Federico Semeraro, Ricardo Leiderman, André M.B. Pereira

An image-based simulation CUDA toolkit for material characterization using Finite Element numerical methods to compute the effective thermal conductivity, elasticity, and permeability.

Publication Github Presentation

SoftwareX 2021

Porous Microstructure Analysis (PuMA)

Authors: Joseph C. Ferguson, Federico Semeraro, John M. Thornton, Francesco Panerai, Arnaud Borner, Nagi N. Mansour

PuMA was developed to compute effective material properties and perform material response simulations on 3D microstrutural images obtained from X-ray tomography.

Publication Github Website

AIAA SciTech 2021

Multi-Scale Analysis of Effective Mechanical Properties of Porous 3D Woven Composite Materials

Authors: Sergio Fraile Izquierdo, Federico Semeraro and Marcos Acín

Computation and analysis of the mechanical properties of 3D woven composite materials that share the same yarn structure but differ in their matrix porosity and yarn's fiber volume fraction.

Publication

Computational Materials Science 2021

Anisotropic analysis of fibrous and woven materials part 2: Computation of effective conductivity

Authors: Federico Semeraro, Joseph C. Ferguson, Marcos Acin, Francesco Panerai, Nagi N. Mansour

Numerical finite volume method to compute the effective thermal conductivity of 3D microstructures obtained from tomography, accounting for the anisotropy of the constituent phases.

Publication Presentation

Computational Materials Science 2020

Anisotropic analysis of fibrous and woven materials part 1: Estimation of local orientation

Authors: Federico Semeraro, Joseph C. Ferguson, Francesco Panerai, Robert J. King, Nagi N. Mansour

Three techniques to estimate the local orientation of fibrous microstructures obtained from tomography scanning: a common image processing technique called structure tensor, a method based on artificial flux, and a novel ray-casting approach

Publication Presentation

University

OMS CS7643 Deep Learning - arXiv 2023

NeRF applied to satellite imagery

Authors: Federico Semeraro, Yi Zhang, Wenying Wu, Patrick Carroll

Surf-NeRF, a method to synthesize novel views from a sparse set of satellite images of a scene as well as its surface elevation, while accounting for the variation in lighting present in the pictures.

Publication Github

OMS CS6476 Computer Vision

Stereo Correspondence using Graph-Cuts

Two techniques to perform stereo correspondence: a simple method based on SSD and a graph cut algorithm. The performance of both method was benchmarked against the Middlebury image database.

Final Report

OMS CS6475 Computational Photography

Video Stabilization using L1 optimal camera paths

An automatic video stabilizer relying on the estimation of the original camera path through the detection of features between video frames and the computation of the similarity transformation between them.

Final Report

Internships

blender2dirsig

Automated the creation of 3D synthetic scenes during the summer internship at Planet Labs (2022) for the DIRSIG image generation model using Blender scripting.

Presentation

Copper Tracker

Developed algorithms during the summer internship at Earth-i (2017) to correlate copper ore extraction data with features observable from space, tracked using Convolutional Neural Networks.

Presentation Article

LiMA Flight Tracker

Developed app using R during the year-long internship in the Data Science department at Airbus UK (2015-2016). The objective was to analyze aircraft movements through FlightRadar24 and predict maintenance stops.

Github Demo

Personal

ESA NEMESYS Bexus

Designed the structure of a student experiment (NEMESYS Bexus) flown on a stratospheric balloon by the European Space Agency to study the effect of particle impacts on memory boards.

Webpage

Water Brigade - Ghana 2014

Joined a brigade organized by Imperial College London traveling to a village in Ghana with no access to potable water, with the aim of designing and implementing water systems to prevent water-related illnesses.

Webpage