Diego Velazquez
AI ML Engineer at Brain Co.
Barcelona, Catalonia, Spain
Summary
🤩
Rockstar🎓
Top SchoolDiego Velazquez is an AI/ML engineer and researcher with nine years of experience building foundation-model-scale systems, from dataset curation and large-scale pretraining to human-in-the-loop evaluation and production deployment. He led design and engineering for Welocalize’s first LLM-based post-editing and translation-quality suite—cutting costs by over 50%, reducing errors by 28%, and winning an AI Breakthrough Award—while also contributing open research like EarthView and EarthMAE during a visiting role with ServiceNow and ElementAI. His work spans vision, language, and multimodal foundation models with strengths in self-supervised learning, robust training, and tooling for agentic workflows and automated evaluation. An active open-source contributor, he has improved core image-transforms and testing in the Kornia library and earlier added high-level APIs and model exports to OpenCV’s ecosystem. Based in Barcelona, he combines PhD-level research rigor with pragmatic product delivery and a knack for turning large heterogeneous datasets into reproducible, deployable models.
9 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Vision, Doctor of Philosophy - PhD, Computer Vision at Universitat Autònoma de Barcelona
Github Skills (7)
computer-vision10
pytorch10
python10
image-processing10
testing9
machine-learning9
neural-network8
Programming languages (9)
Github contributions (5)
kornia/kornia
May 2019 - Jan 2020
🐍 Geometric Computer Vision Library for Spatial AI
Role in this project:
ML Engineer
ML EngineerContributions:182 commits, 26 PRs, 14 pushes in 9 months
Contributions summary:Diego implemented and tested image normalization functionality within the Kornia library, focusing on channel-wise mean and standard deviation normalization. They added new classes and functions for image transformations and color space conversions, including RGB to BGR, BGR to RGB, and RGB to HSV. Further contributions involved adding tests to the image module, which involved correcting various errors in the tests and ensuring that the test cases for the implemented functions work as expected.
pytorchdifferentiablepythonvisiondeep-learning
dvd42/opencv
Jun 2019 - Feb 2020
Open Source Computer Vision Library
Contributions:8 PRs, 147 pushes, 24 branches in 8 months
image-processingpythoncomputer-visionvision
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.