Alp Kucukelbir
Director, AI, Supply Chain at Fero Labs
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolAlp Kucukelbir is a data-driven AI leader and entrepreneur with 13 years of experience building probabilistic modeling and optimization systems across academia and industry. He co-founded and serves as Chief Scientist at Fero Labs, teaches as an adjunct professor at Columbia, and now leads AI for supply chain at Amazon, bridging research rigor with product impact. His PhD from Yale and postdoctoral work at Princeton/Columbia underpin contributions to core probabilistic programming projects—most notably variational Bayesian inference code for Stan and implementation of reparameterization gradients in Edward. Alp blends hands-on algorithm development with strategic leadership, translating complex Bayesian methods into production-ready systems for decision-making under uncertainty. He also has a track record of mentoring students and founding teams, and an uncommon knack for turning advanced inference techniques into scalable supply-chain solutions.
13 years of coding experience
PhD Engineering and Applied Science, PhD Engineering and Applied Science at Yale UniversityB.A.Sc. (Hon) Electrical Engineering, B.A.Sc. (Hon) Electrical Engineering at University of Toronto
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (18)
algorithm10
algorithms10
variational-inference10
probabilistic-programming10
c-language10
python10
machine-learning10
implement10
mathematical-optimization10
bayesian10
eigen10
tensorflow10
cprogramming-language10
bayesian-inference10
model-driven9
Programming languages (7)
Github contributions (5)
blei-lab/edward
Mar 2016 - Jul 2017
A probabilistic programming language in TensorFlow. Deep generative models, variational inference.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:243 commits, 42 PRs, 120 pushes in 1 year 4 months
Contributions summary:Alp implemented a `sample` function to the `MFGaussian` class and then added a new class called `KLpq` that appears to be involved with the KL divergence. Moreover, they added the code related to the reparameterization gradient estimation, demonstrating the user's work on implementing the loss function in a variational inference setting. The user has also touched upon implementation of the distribution functions and their connection with the tensor flow.
inferenceneural-networksmachine-learningprobabilistic-programmingdeep-generative-models
stan-dev/stan
Jul 2014 - Apr 2016
Stan development repository. The master branch contains the current release. The develop branch contains the latest stable development. See the Developer Process Wiki for details.
Role in this project:Back-end Developer
Back-end DeveloperContributions:248 commits, 10 PRs, 135 pushes in 1 year 9 months
Contributions summary:Alp made several commits focused on developing the variational Bayesian inference (VB) portion of the Stan library. These changes included implementing and testing ELBO (Evidence Lower Bound) calculations, creating and modifying classes for variational parameters (fullrank and meanfield), and integrating stochastic gradient ascent with adaptive stepsize. The commits demonstrate expertise in probabilistic programming, mathematical optimization, and algorithm implementation.
bayesian-inferencestanmasterbayesian-statisticsdetails
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.