Tingting Markstrum
Software Engineer at Facebook
New York City Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolTingting Markstrum is a software engineer based in the New York City area with five years of engineering experience and a Ph.D. in Information Technology from MIT. Currently at Facebook, she brings a strong background in high-performance and large-scale systems from prior senior roles at Two Sigma and Oracle. Her open-source contributions to Facebook Research’s fairscale show hands-on expertise in ML infrastructure—implementing a DynamicLossScaler and improving FSDP and reduce-scatter performance and correctness. She combines research-level rigor with production-grade engineering, adding type annotations, tests, and clear documentation to complex codebases. Former internships at IBM and SAP reflect an early and consistent focus on systems and research environments. Colleagues describe her as a pragmatic problem-solver who bridges deep algorithmic understanding with reliable, test-driven delivery.
5 years of coding experience
12 years of employment as a software developer
Ph.D, Information Technology, Ph.D, Information Technology at Massachusetts Institute of Technology
Github Skills (8)
pytorch10
machine-learning10
distributed-training10
sdp10
python10
optimizers9
optimizer9
testing8
Programming languages (1)
Github contributions (5)
facebookresearch/fairscale
May 2021 - Mar 2022
PyTorch extensions for high performance and large scale training.
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 67 reviews, 10 commits in 10 months
Contributions summary:Tingting implemented and tested a `DynamicLossScaler` class to support optimizer updates on the CPU and prevent gradient underflow/overflow. They added type annotations, comments, and tests to the codebase. Further contributions include adding and modifying features related to Fully Sharded Data Parallel (FSDP) and reduce scatter operations, including fixing bugs and enabling reduce scatter overlap.
pytorchdeep-learningmachine-learningscaletraining
tmarkstrum/fairscale
Apr 2021 - Apr 2021
PyTorch extensions for high performance and large scale training.
Contributions:2 pushes, 1 branch in 1 day
pytorchdeep-learningmachine-learningscaletraining
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.