Philipp Witte
Member Of Technical Staff at Microsoft AI
Seattle, Washington, United States
Summary
👤
Senior🎓
Top SchoolPhilipp Witte is a Member of Technical Staff at Microsoft AI with 11 years of experience building large-scale training infrastructure, specializing in parallelism, checkpointing, and pre-training systems for MSI models. He previously spent several years as a researcher at Microsoft Research working on physics-informed ML and inverse problems, blending deep learning with high-performance computing. His academic background (PhD work at Georgia Tech and UBC in computational science and geophysics) underpins a pragmatic approach to numerics and scalable ML systems. An active contributor to open-source scientific tooling, he improved Dask-based tutorials in the Devito finite-difference DSL, demonstrating attention to reproducible, parallel workflows. Based in Seattle, he combines research rigor with production-focused engineering to accelerate large model training pipelines.
11 years of coding experience
9 years of employment as a software developer
Master of Science - MS, Geophysics, Master of Science - MS, Geophysics at University of HamburgPh.D., Computational Science and Engineering, Ph.D., Computational Science and Engineering at Georgia Institute of TechnologyPh.D. Candidate, Geophysics, Ph.D. Candidate, Geophysics at The University of British Columbia
Github Skills (8)
jupyter-notebook10
dask10
python10
dsl9
finite-difference9
compiler7
code-generation7
compiler-compiler7
Programming languages (4)
Github contributions (5)
devitocodes/devito
Feb 2020 - Mar 2020
DSL and compiler framework for automated finite-differences and stencil computation
Role in this project:
Data Scientist
Data ScientistContributions:13 commits, 2 PRs, 7 pushes in 7 days
Contributions summary:Philipp focused on cleaning up and improving a Dask tutorial within the Devito repository. This included modifying the content of the tutorial's notebook, specifically addressing the implementation of parallel forward modeling and FWI objective functions using Dask. The changes involved ignoring certain outputs and merging branches, indicating a focus on maintaining the tutorial's functionality and usability.
code-generationautomatic-differentiationfinite-differencecompilerultrasound-imaging
slimgroup/Azure2019
Aug 2019 - Aug 2020
Azure workflow for SLIM algorithms.
Contributions:1 release, 80 commits, 4 PRs in 1 year
workflowslimazure
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.