Francesco Pochetti
Staff Machine Learning Engineer ML Team Lead at Bolt
Brussels, Belgium
Summary
👤
Senior🎓
Top SchoolFrancesco Pochetti is a Staff Machine Learning Engineer and ML Team Lead based in Brussels with 12 years of experience applying ML and computer vision to real-world product problems, currently leading the user identity domain at Bolt. He progressed from data scientist roles at Amazon and Bolt to building and managing end-to-end identity verification stacks—spanning OCR, liveness, face matching and authentication—while balancing hands-on engineering with team leadership. His background as a chemist-turned-ML practitioner informs a disciplined experimental approach and strong telemetry-driven decision making. Francesco contributes to open-source computer vision tooling (notably improvements to the airctic/icevision framework) and blogs about his ML/DL journey, signaling a commitment to knowledge sharing. He has deep production ML experience across cloud platforms, pipelines and model monitoring, plus a track record of introducing tooling and best practices to teams.
12 years of coding experience
7 years of employment as a software developer
Master's degree Physical-Chemistry; Chemistry; Physical and Natural Sciences; Inorganic Phisycal-Chemistry, Master's degree Physical-Chemistry; Chemistry; Physical and Natural Sciences; Inorganic Phisycal-Chemistry at Sapienza Università di Roma
English, French, Italian
Stackoverflow
66reputation
5kreached
1answer
6questions
Github Skills (10)
object-detection10
computer-vision10
pytorch10
fastai10
deep-learning10
python10
pytorch-lightning9
huggingface-transformers6
huggingface-tokenizers6
amazon-sagemaker6
Programming languages (6)
Github contributions (5)
airctic/icevision
Oct 2020 - Mar 2022
An Agnostic Computer Vision Framework - Pluggable to any Training Library: Fastai, Pytorch-Lightning with more to come
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:1 release, 95 reviews, 74 commits in 1 year 5 months
Contributions summary:Francesco contributed to several aspects of the `airctic/icevision` repository, including refactoring code, fixing bugs, and adding new functionalities. The user updated existing parsing methods by switching to `pathlib` and updated the code to accomodate for issues in `albumentations`. They also implemented features such as saving/loading records from pickle and added functionality for showing actuals in predictions, and integration of Keypoint functionality.
voc-parservision-frameworkpythonagnostictraining
FraPochetti/KagglePlaygrounds
Aug 2018 - Dec 2022
Contributions:56 commits, 4 PRs, 147 pushes in 4 years 4 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.