Arkanath Pathak
Senior Software Engineer (ML Theory) at Google
Irvine, California, United States
Summary
👤
Senior🎓
Top SchoolArkanath Pathak is a Senior Software Engineer with 11 years of experience at Google and DeepMind, specializing in machine learning theory, multimodal systems, and robust ensemble methods. His work spans research and production: from Brand Safety and video prediction at YouTube to developing multimodal LLM distillation and image-context tools at DeepMind/Google Research. He contributes to influential open-source projects like TensorFlow’s AdaNet, adding ensemble strategies, mean ensembling, and subnetwork export features that bridge theory and practical AutoML. Based in Irvine, California, he combines deep academic training from IIT Kharagpur with hands-on systems engineering to tackle distribution shift, model scaling, and detection of generative-model outputs. A less obvious strength is his track record of moving research ideas into scalable tooling that supports evaluation and deployment across large platforms.
11 years of coding experience
1 year of employment as a software developer
Bachelor's (B.Tech.) Computer Science, Bachelor's (B.Tech.) Computer Science at Indian Institute of Technology, Kharagpur
Hindi, English
Stackoverflow
42reputation
7kreached
2answers
4questions
Github Skills (16)
machine-learning10
deeplearning-ai10
deep-learning10
tensorflow10
automl10
python10
graph6
java6
max-heap6
algorithm6
data-structures6
magicsquare6
adjacency-matrix6
dynamic-programming6
mysql6
Programming languages (2)
Github contributions (5)
tensorflow/adanet
Jan 2019 - May 2020
Fast and flexible AutoML with learning guarantees.
Role in this project:
ML Engineer
ML EngineerContributions:9 commits in 1 year 4 months
Contributions summary:Arkanath primarily contributes to the AdaNet library, focusing on ensemble methods. Their work involves defining and implementing various ensemble strategies, including grow, all, and complexity-regularized approaches. They also added support for mean ensembling and exporting subnetworks' logits and last layers. Furthermore, the user fixed and modified the code related to evaluation metrics.
automlmachine-learningflexibledeep-learningdistributed-training
arkanath/ds
Apr 2016 - Apr 2016
Contributions:49 commits in 7 days
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.