Steven Bradtke
Senor Engineer at Maryland Research Institute
United States
Summary
👤
Senior🎓
Top SchoolSteven Bradtke is a senior engineer and research leader with a PhD in Computer Science and over four decades of cumulative technical experience, including eight years in his current focused roles. He specializes in AI and machine learning—particularly reinforcement learning, neural sequence models, and NLP—and has practical experience contributing to prominent open-source projects like AWS Sockeye for neural machine translation. Steven blends deep research credentials with hands-on engineering, having led algorithm design, real-time control, and large-scale distributed computation across government and industry labs. He frequently serves as a technical lead, mentoring teams while shipping robust production features such as learning-rate scheduler persistence and checkpoint-driven training controls. Based in the United States, he pairs academic rigor with pragmatic problem solving across diverse application domains and legacy-to-modern system integrations. An understated strength is his ability to modernize training workflows and reproducibility in ML systems, improving long-running experiment reliability.
8 years of coding experience
19 years of employment as a software developer
Masters of Science Computer Science, Masters of Science Computer Science at University of MichiganPhD Computer Science, PhD Computer Science at University of Massachusetts AmherstBachelor of Science Computer Science, Bachelor of Science Computer Science at Michigan State University
Github Skills (9)
sequence-to-sequence10
machine-translation10
pytorch10
deep-learning10
neural-machine-translation10
python10
machine-learning9
attention-mechanism9
tensorflow3
Programming languages (2)
Github contributions (5)
awslabs/sockeye
Mar 2019 - Feb 2021
Sequence-to-sequence framework with a focus on Neural Machine Translation based on PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:15 reviews, 5 commits, 5 PRs in 1 year 11 months
Contributions summary:Steven contributed to the `sockeye` project, a sequence-to-sequence framework focused on neural machine translation. Their work included adding functionality to stop training after a specified number of checkpoints and fixing a print statement in a learning rate scheduler. The user also implemented saving and restoring the learning rate scheduler and updated the code to properly handle learning rate warmup during continued training. Furthermore, they added a feature to cache "best" parameter sets to a subdirectory.
sequencetranslationsockeyeneural-machine-translationseq2seq
tuglat/sockeye-scripts
Mar 2019 - May 2019
Scripts and support utilities for Sockeye
Contributions:3 pushes in 2 months
sockeye
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.