Bryan Perozzi

New York, New York, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Bryan Perozzi is a researcher-practitioner who builds scalable neural methods for learning expressive representations of social relationships and natural language, with 11 years of industry experience and a strong academic track record of 20+ peer-reviewed papers at top ML and data mining venues. He designs algorithms used for prediction, pattern discovery, and anomaly detection on large graph-structured data and has translated that research into production-grade systems. Bryan is an active open-source contributor—most notably enhancing the influential DeepWalk project for graph embeddings and improving graph sampling and RDF support in TensorFlow GNN—demonstrating a focus on usability and scalability. Based in New York, he blends academic rigor with practical engineering, having worked in data science before the term was mainstream. He favors solutions that scale in memory and compute, evidenced by work on parallelized walk generation and optimized sampling pipelines. Bryan brings a rare combination of deep theoretical insight and hands-on system-building to complex networked-data problems.

11 years of coding experience

Stony Brook University

Github Skills (18)

graph-algorithms10

python10

gnn10

machine-learning10

sampling10

parallel-processing10

tensorflow10

sampler10

graph-database9

word2vec9

gensim9

graph9

graph-datastructures9

algorithms8

algorithm8

Programming languages (2)

Jupyter NotebookPython

Github contributions (5)

phanein/deepwalk

Aug 2014 - Mar 2016

DeepWalk - Deep Learning for Graphs

Role in this project:

Back-end Developer

Contributions:13 commits, 2 PRs, 2 pushes in 1 year 7 months

Contributions summary:Bryan significantly enhanced the functionality of the DeepWalk project, focusing on improvements to its core features and usability. They refactored the command-line interface and expanded the input file format support for the graph processing module. Additionally, the user implemented a feature allowing the saving of generated walks to disk, optimizing memory management for large graphs. Finally, they contributed to parallelizing walk generation and included a scoring routine example.

pytorchdeep-learning-for-graphsdeep-learningmachine-learninggraph-neural-networks

tensorflow/gnn

Sep 2021 - Dec 2022

TensorFlow GNN is a library to build Graph Neural Networks on the TensorFlow platform.

Role in this project:

ML Engineer

Contributions:21 commits, 1 push, 15 comments in 1 year 2 months

Contributions summary:Bryan primarily contributed to the TensorFlow GNN library by modifying the graph sampling components. Their work involved implementing a uniform random sampling strategy, enhancing existing sampling methods, and improving the validation checks and error messages related to feature sizes. The changes also included updates to the sampling spec proto definition and adjustments to the schema augmentations, demonstrating a focus on improving the flexibility and user-friendliness of the sampling pipeline. The user also added a triple converter for RDF-style input.

gnndeep-learningneural-graphneural-networksmachine-learning

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.

Request Free Trial