Edouard Mehlman
Head Of Machine Learning at Feedly
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolEdouard Mehlman is Head of Machine Learning at Feedly with a decade of experience building production ML and NLP systems and leading small engineering teams. He blends research-quality expertise in variational inference, normalizing flows and deep generative models—developed at Berkeley and ENS—with hands-on engineering: backend development, test automation, deployment and profiling. His open-source contributions include improving test suites and architecture for scvi-tools, a well-regarded library for probabilistic single-cell omics analysis, reflecting a strong focus on reliability and reproducibility. Comfortable bridging academia and product, he has delivered anomaly-detection and time-series solutions in industry research settings and scaled ML features for content understanding at Feedly. A former physics teacher and cadet officer, he brings disciplined mentorship and clear communication to complex technical challenges.
10 years of coding experience
1 year of employment as a software developer
Computer Science Applied Mathematics Economics, Computer Science Applied Mathematics Economics at École PolytechniqueLycée Sainte-GenevièveMVA Master of Science M2. Mathématiques Vision Apprentissage, MVA Master of Science M2. Mathématiques Vision Apprentissage at École Normale Supérieure Paris-Saclay
French, English, Spanish
Github Skills (6)
unit-testing10
pytorch10
python10
test-automation10
machine-learning9
variational-autoencoder9
Programming languages (6)
Github contributions (5)
scverse/scvi-tools
Apr 2018 - Jan 2019
Deep probabilistic analysis of single-cell and spatial omics data
Role in this project:
Backend Developer & Test Automation Engineer
Backend Developer & Test Automation EngineerContributions:161 commits, 35 PRs, 77 pushes in 9 months
Contributions summary:Edouard's contributions centered on enhancing the test suite and improving the project's structure. They refactored test-related logic, separated training and dataset components, and added options for dataset dropout and imputation tests. The user also worked on supporting the inclusion of a Travis build for the project.
hierarchicaldeep-generative-modelmixture-of-expertssingle-cell-rna-seqsingle-cell
Edouard360/webgl-mark2
Jul 2016 - Sep 2016
Contributions:91 commits, 1 PR, 53 pushes in 1 month
benchmarkrenderingrendering-enginewebgl
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.