Available for collaborations · Barcelona, Spain

Víctor Escribano

AI Engineer

I build computer vision and deep learning systems for robotics and real-world perception — from autonomous vehicles and agricultural robots to real-time people tracking and foveated small-object detection.

Víctor Escribano García
5+ years building
vision systems
01

About me

I'm an Industrial Electronics & Automatic Control Engineer with an MSc in Robotics and Automatic Control from the Polytechnic University of Catalonia (UPC). Today I work as an AI Engineer at Escribano Mechanical & Engineering, building real-time tracking and deep learning perception for autonomous robotic systems.

My path runs through autonomous racing, agricultural robots, astrophysics instrumentation and award-winning hackathons. The common thread: turning cameras, sensors and neural networks into systems that perceive and act in the real world.

“There may be no better way to communicate my passion and expertise than through the projects showcased below. Each one represents a unique problem solved and a step forward in my journey as an engineer.”
MSc Robotics & Automatic Control · UPC
🏆 1st HackUPC 2023 — Overall Winners
🥈 2nd Best autonomous car · Formula Student Spain 2020
🎓 Honors Final thesis — Inverse Perspective Mapping
02

Experience

AI Engineer

Jun 2025 — Present

Escribano Mechanical & Engineering (EM&E) · Barcelona

  • Built a real-time "follow-me" person-following system for autonomous robotic platforms — detecting, tracking and re-identifying people on a CPU-only embedded device, fusing LiDAR and camera and processing point clouds in real time.
  • Designing new deep learning models — such as hyperbolic foveation imaging to detect small objects — that focus high-resolution sampling on the region of interest while cutting peripheral compute.
  • Deploying and optimizing deep learning inference on embedded edge hardware for low-latency on-device perception.
PythonPyTorchOpenVINOObject TrackingRe-IDLiDAR–Camera FusionPoint CloudsEmbedded (CPU)ROS2

Robotics Perception Engineer

Feb 2024 — Jun 2025

Earth Rover

  • Built perception systems for autonomous navigation in ground-based agricultural robots.
  • Enhanced computer vision models using synthetic data and generative-AI augmentations.
C++ROSGazeboPoint CloudsBlenderYOLOStable Diffusion

QA Automation Engineer

Feb 2023 — Jan 2024

Bertrandt

  • Developed Python automation for protocol trace analysis and validation under SCRUM.
  • Designed test catalogs from client requirement specifications.
PythonSeleniumFastAPIOpenCVCANoePyAutoGUI

Software Engineer (Scholarship)

Jul 2022 — Oct 2024

Canary Islands Institute of Astrophysics (IAC)

  • Developed a hardware-agnostic telescope pointing estimator to minimize ephemerides waiting times, accelerated with Numba.
  • Collaborated with Military Defense on detecting and tracking satellites and space debris.
  • Upgraded the IAC80 & TCS telescope control systems with a pyQt + ROS1 real-time interface.
PythonNumbapyQtROS1Space Surveillance

Driverless Perception Engineer

Jan 2020 — Sep 2022

UPC ecoRacing · Formula Student

  • Designed the perception system (camera + LiDAR) of UPC's first autonomous racing vehicle.
  • Implemented mono & stereo vision, real-time object classification and sensor fusion.
  • 2nd best autonomous racing car in Formula Student Spain 2020.
PythonC++ROS1OpenCVYOLOCUDATensorRTLiDAR

Robotics Specialist (Internship)

Sep 2021 — Jan 2022

Alisys Robotics

  • Built real-time face recognition systems for mobile and medical robots.
  • Studied and documented an omni-transportation platform for medical robots.
OpenCVROS1FlaskTkinterMATLAB
03

Selected Projects

TelescopePointingEstimator at the IAC

TelescopePointingEstimator — IAC

Modernized the Teide Observatory night-telescope control systems (ROS + pyQt) and built a hardware-agnostic, Numba-accelerated pointing estimator that maximizes satellite captures per night — 0.9 s average wait, ~3.2 images/min.

PythonNumbaROSpyQtAstronomy
Synthetic crack dataset

Synthetic Crack Dataset — Blender

A Blender-based synthetic pipe-crack generator producing diverse, automatically labeled images for advanced segmentation and YOLOv8 training — tested on real images with strong results on challenging scenarios.

BlenderYOLOv8Synthetic DataSegmentation
IPM GAN top-view generation

IPM GAN

Generating a top-view map (Inverse Perspective Mapping) directly from a front camera view using Generative Adversarial Networks.

GANPyTorchIPM
HackUPC 2023 winners

HackUPC 2023 — 🏆 Winners

Overall winners in under 36h: real-estate interior remodeling fusing the Segment Anything Model, the Restb.ai API and Stable Diffusion, with a VR/Blender spherical-projection UI.

SAMStable DiffusionVRRestb.ai
UPC ecoRacing first autonomous car

ecoRD 2020/21 — Driverless

Camera + LiDAR perception for UPC ecoRacing's first autonomous vehicle, leading to the 2nd best autonomous car at Formula Student Spain 2020/21.

Sensor FusionLiDARROSTensorRT
UPC ecoRacing second-generation autonomous car

ecoRD 2021/22 — Driverless

A camera-based perception system for UPC ecoRacing's second-generation autonomous car during the 2021/22 season.

Computer VisionYOLOROSTensorRT
Frontier exploration in ROS2

Last Frontier Exploration

A personal improvement on closest-frontier autonomous exploration in ROS2 — mapping and navigating unknown indoor environments.

ROS2SLAMAutonomy
Tennis movement analysis

Tennis Movement Analysis

Inspired by the US Open final (Alcaraz vs. Ruud): YOLO detects and classifies both players, and planar homographies project a live top-view court to visualize their movement heatmaps.

YOLOHomographyOpenCV
04

Skills & Tools

Languages

PythonC++MATLAB

AI / Computer Vision

PyTorchTensorFlowOpenCVYOLOGANsStable DiffusionSAMObject TrackingRe-ID

Robotics

ROS1ROS2GazeboSensor FusionLiDARPoint Clouds

Deploy & Acceleration

OpenVINOTensorRTCUDANumbaNVIDIA JetsonLinuxGit

Languages (spoken)

Spanish — nativeCatalan — nativeEnglish — B2

Also

BlenderFastAPIpyQt3D Printing
05

Let's build something that sees.

Open to AI / computer-vision & robotics opportunities and collaborations.

vesgarcia99@gmail.com