Edoardo Conti
Senior Machine Learning Engineer at Databricks
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolEdoardo Conti is a Senior Machine Learning Engineer with 12 years of experience building large-scale ML systems for Uber, Meta (where he co-founded the applied reinforcement learning team and contributed to the widely used ReAgent project), and startups, and now optimizing systems and infrastructure at Databricks. As a YC founder (Bandit ML) he applied bandit algorithms to real-world recommendation and promotion problems for customers like Namecheap, then led an acquisition into Silo, giving him product intuition alongside engineering depth. He pairs hands-on implementation—evidenced by contributions that enhanced discrete action prediction and batch-constrained Q-learning in ReAgent—with leadership roles as an engineering manager and Sequoia scout. Based in San Francisco, he combines quantitative training from MIT and Rutgers with operator experience to move research into production, and is an active angel investor spotting early-stage ML opportunities.
12 years of coding experience
11 years of employment as a software developer
S20, S20 at Y CombinatorMaster's Degree Finance (quant), Master's Degree Finance (quant) at Massachusetts Institute of TechnologyBachelor's degree Math Computer Science, Bachelor's degree Math Computer Science at Rutgers University
Stackoverflow
21reputation
1kreached
0answers
2questions
Github Skills (9)
pytorch10
machine-learning10
reinforcement-learning10
python9
deep-learning9
caffe8
char6
realloc6
pointer6
Programming languages (5)
Github contributions (5)
facebookresearch/ReAgent
Mar 2018 - Jun 2019
A platform for Reasoning systems (Reinforcement Learning, Contextual Bandits, etc.)
Role in this project:
ML Engineer
ML EngineerContributions:125 commits, 13 PRs, 2 pushes in 1 year 2 months
Contributions summary:Edoardo implemented and refined components of a discrete action predictor within the reinforcement learning platform. They modified the code to return both max q and softmax policies from the discrete action model, and ensured the test suite passed with either policy. Further modifications included enhancements for action selection in parametric models. Additional contributions were focused on supporting batch-constrained Q-learning by adding related features.
reinforcement-learningcontextualbanditscontextual-banditsreinforcement
banditml/offline-policy-evaluation
Mar 2020 - Jul 2021
Implementations and examples of common offline policy evaluation methods in Python.
Contributions:4 releases, 12 reviews, 132 commits in 1 year 4 months
policyimplementationscounterfactual-policy-evaluationoffline-policy-evaluationpython
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.