Ihor Indyk
Software Engineer at Google
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolIhor Indyk is a Software Engineer at Google with nine years of experience blending adversarial machine learning research and production ML infrastructure work. He holds a PhD in Pure and Applied Mathematics and researches defenses for Support Vector Machines while contributing to large open-source projects like TensorFlow TFX and Apache Beam to optimize data transformations and reduce memory usage. At Google he has moved ideas from research to production—implementing parquet outputs and pyarrow.RecordBatch support in TFX and improving ApproximateQuantiles performance in Beam—demonstrating a rare mix of theoretical rigor and systems engineering. Based in San Francisco, he combines a strong statistics background with hands-on backend development to make ML pipelines more efficient and robust.
9 years of coding experience
1 year of employment as a software developer
Master of Science - MS, Statistics, A, Master of Science - MS, Statistics, A at Kiev National Taras Shevchenko UniversityDoctor of Philosophy - PhD, Pure and Applied Mathematics, Doctor of Philosophy - PhD, Pure and Applied Mathematics at Stevens Institute of Technology
Github Skills (21)
data-pipelines10
python10
memory-management10
big-data10
machine-learning10
tensorflow10
performance-optimization10
data-pipeline10
apache-beam10
data-structure9
algorithm9
algorithms9
mlops9
data-structures9
parquet8
Programming languages (3)
Github contributions (5)
tensorflow/tfx
Dec 2020 - Dec 2022
TFX is an end-to-end platform for deploying production ML pipelines
Role in this project:
ML Engineer
ML EngineerContributions:7 reviews, 13 commits, 24 comments in 2 years
Contributions summary:Ihor's contributions primarily revolve around enhancing the TFX platform for production machine learning pipelines. They focused on benchmarking the TFX Transform component, adding metrics for analysis cache optimization and output materialization. Furthermore, they implemented features like using pyarrow.RecordBatch outputs and outputting parquet files, improving the efficiency and flexibility of the transformation process within TFX. These changes reflect a focus on optimizing data handling and integration within the machine learning workflow.
deployingend-to-endml-pipelinesmlmlops
apache/beam
Aug 2020 - Sep 2021
Apache Beam is a unified programming model for Batch and Streaming data processing.
Role in this project:Back-end Developer
Back-end DeveloperContributions:34 reviews, 5 commits, 4 PRs in 1 year 1 month
Contributions summary:Ihor focused on enhancing the `ApproximateQuantiles` transform within the Apache Beam framework, specifically addressing functionality related to non-uniform weights. Their contributions involved modifying code in `sdks/python/apache_beam/transforms/stats.py` to improve performance and memory usage, including batched merging of accumulators for increased efficiency. Additionally, the user made changes related to `TupleCombineFn` to decrease peak memory usage.
golangpythonstreaming-databeambatch
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.