Enrico Liscio
Postdoctoral Researcher at Delft University of Technology
Rijswijk, South Holland, Netherlands
Summary
🤩
Rockstar🎓
Top SchoolEnrico Liscio is a postdoctoral researcher and NLP specialist with eight years of experience bridging deep learning research and industrial computer vision. His work combines rigorous AI ethics and value inference—stemming from a Cum Laude PhD in AI—with hands-on model engineering, exemplified by concrete contributions to the popular keras-retinanet codebase where he improved loss separation and anchor preprocessing. Based at Delft and active in the Algosoc Gravitation Program, he builds hybrid intelligence methods to infer context-specific human values and make AI systems more aligned with real people. Previously he led computer vision projects for automation and logistics as a technical lead, shipping production-ready deep learning solutions for clients like ABB and Vanderlande. Colleagues describe him as equally comfortable in code, experiments, and the philosophical questions behind AI behavior. He brings a rare combination of top-tier academic credentials, production ML experience, and a focus on making AI understand human values.
8 years of coding experience
6 years of employment as a software developer
University of BolognaTU DelftCum Laude MSc degree, Systems and Control, 9/10, Cum Laude MSc degree, Systems and Control, 9/10 at Delft University of Technology
Italian, English, French, Dutch
Github Skills (10)
retinanet10
object-detection10
keras10
computer-vision10
machine-learning10
loss-functions10
trainings10
python10
modeling10
tensorflow9
Programming languages (3)
Github contributions (5)
fizyr/keras-retinanet
Oct 2017 - Oct 2018
Keras implementation of RetinaNet object detection.
Role in this project:
ML Engineer
ML EngineerContributions:15 commits, 9 PRs, 12 pushes in 1 year
Contributions summary:Enrico made several modifications related to training and loss functions within the RetinaNet model, specifically separating classification and regression losses. They adjusted the model architecture by changing kernel sizes and tracked losses separately, as well as modified pre-processing to include anchor target generation. These changes involved modifying the model's architecture, loss calculation, and data preparation steps to enhance its performance and functionality.
deep-learningretinanetcomputer-visionobject-detectiontensorflow
enricoliscio/enricoliscio.github.io
Mar 2021 - Feb 2025
Contributions:99 pushes, 1 branch in 3 years 11 months
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.