Nicki Detlefsen
Associate Professor
Køge, Region Zealand, Denmark
Summary
🤩
Rockstar🎓
Top SchoolNicki Detlefsen is an Associate Professor at DTU with nine years of experience at the intersection of machine learning research, MLOps, and software engineering. With a PhD in Mathematics and Computer Science, she moved from postdoctoral work on deep generative models and manifold learning into building reliable ML infrastructure and teaching MLOps at DTU. She contributes to PyTorch Lightning as a software engineer, improving core training utilities and removing brittle dependencies to make large-model training more robust and portable. Her GitHub work includes course material and hands-on VAE implementations that emphasize profiling and performance tuning, reflecting a pragmatic focus on efficient model training. Known for bridging rigorous research and production-grade tooling, she mentors contributors and students while shaping reproducible ML workflows.
9 years of coding experience
4 years of employment as a software developer
Technical University of Denmark
Danish, English
Github Skills (18)
pytorch10
python10
machine-learning10
pytorch-lightning10
deep-learning10
trainings10
model-optimization10
modeling10
image-processing9
evaluation9
computer-engineering9
eval9
apidoc8
api8
tensorboard8
Programming languages (6)
Github contributions (5)
SkafteNicki/dtu_mlops
Apr 2021 - Jan 2023
Exercises and supplementary material for the machine learning operations course at DTU.
Role in this project:
ML Engineer
ML EngineerContributions:112 reviews, 570 commits, 339 PRs in 1 year 9 months
Contributions summary:Nicki's commits primarily focus on developing and experimenting with a Variational Autoencoder (VAE) model for MNIST image reconstruction. They implement the encoder, decoder, and loss function using PyTorch. Furthermore, they integrate the PyTorch profiler to analyze the model's performance and optimize training. They also demonstrate expertise by using the training loop and model evaluation and saving weights.
pythondtudata-sciencesupplementary-materialmachine-learning
Lightning-AI/pytorch-lightning
Jan 2020 - Jan 2023
Pretrain, finetune ANY AI model of ANY size on multiple GPUs, TPUs with zero code changes.
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:871 reviews, 97 commits, 157 PRs in 2 years 11 months
Contributions summary:Nicki primarily contributed to improving the functionality and robustness of the PyTorch Lightning library. They removed a dependency on the pandas library, refactored the existing code to use a generic csv library, and added additional checks. They also addressed deprecated arguments in learning rate step functions and implemented a fix for ReduceLROnPlateau, demonstrating expertise in optimizing training procedures. These changes showcase skills relevant for back-end development and experience in building and maintaining machine learning models.
pythonheadachespytorch-modelsdata-sciencehandling
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.