robin.schmidt.97@web.

de �
Robin M. Schmidt linkedin.com/schmidt-robin �
robinschmidt.netlify.com �
github.com/SirRob1997 �
Experience Robin M. Schmidt �

Apple Jul 2021 – Present

Machine Learning Research Engineer New York City, USA
◦ Member of the Machine Translation team that is responsible for translations across the Apple ecosystem [Python, Bash]
(many billion words/day) with requests coming from the “Translate” app, Safari, Siri, Camera, etc.
◦ Improved and productionized multilingual machine translation models that boosted translation quality
up to 5 BLEU across many language pairs and reduced the number of deployed models.
AI/ML Resident
◦ Selected among over 8,000 applicants for the first global cohort of 10 AI/ML Residents at Apple.
◦ Advanced the translation accuracy, inference speed, and memory usage of the Transformer models by
implementing novel architectures including research on non-autoregressive approaches.

Max Planck Society (MPI-IS & MPI-INF) Oct 2019 – Mar 2021
Machine Learning Researcher Tübingen, Germany
◦ Advanced development of the “DeepOBS” optimization framework by fixing and developing key features [Python, Bash]
that included batch comparison scripts, analysis code, and multiple endpoints.
� Published an empirical analysis of 15 stochastic non-convex optimization methods on 8 problems result-
ing in 50,000 individual training runs at ICML 2021 (supervisor: Prof. Philipp Hennig).
� Developed novel Domain Generalization methods for CNN-based image classification that led to state-
of-the-art advancements on various generalization datasets (supervisor: Prof. Zeynep Akata).

IBM Aug 2019 – Oct 2019

Software Engineer Intern (Backend) Böblingen, Germany
◦ Lead Extreme Blue intern team and conceptualized more effective instance-level master data graph rep- [Python, JS]
resentations that opened up new use-case sectors for Internet of Things requirements.
◦ Developed a prototype with React.js, a Python RESTful API, and Cassandra & JanusGraph databases
which allowed users to connect and visualize sensor data in real-time.
◦ Presented the results to the team’s global head and wrote requested summary detailing value proposition
and implementation details for senior leadership and offering management.

Eisenmann SE Oct 2015 – Oct 2018

Software Engineer (Backend) Böblingen, Germany
◦ Conceptualized and implemented an uncertainty-based single product tracking system that significantly [Java, JS]
improved transparency and insights for production lines with limited sensor data.
◦ Improved the configurable data analysis pipeline for the Manufacturing Execution System “E-MES” by
implementing better reporting-frameworks for convenient customer usage.
◦ Took the initiative to solve under-specified sensor data collection problems on-premise for a customer
in the USA, which was essential for the team to successfully complete the project on time.

Education

University of Tübingen, Germany Oct 2018 – Mar 2021

Master of Science (M. Sc.) in Computer Science GPA: 1.4/1.0 (Germany) | 3.8/4.0 (USA)

Cooperative State University Stuttgart, Germany Oct 2015 – Oct 2018

Bachelor of Science (B. Sc.) in Computer Science GPA: 2.0/1.0 (Germany) | 3.3/4.0 (USA)

Peer-Reviewed Publications
2022 [1] Robin M. Schmidt, et al. 2022. “Non-Autoregressive Neural Machine Translation: A Call for Clarity”. Proceedings of
the 2022 Conference on Empirical Methods in Natural Language Processing, EMNLP, (under submission).
2021 [2] Kaustubh D. Dhole, […], Robin M. Schmidt, et al. 2021. “NL-Augmenter: A Framework for Task-Sensitive Natural
Language Augmentation”. Proceedings of Anonymous Machine Learning Conference, (under submission).
 [3] Robin M. Schmidt, Frank Schneider, and Philipp Hennig. 2021. “Descending through a Crowded Valley –
Benchmarking Deep Learning Optimizers”. Proceedings of the 38th International Conference on Machine Learning,
ICML, (acceptance rate: 21.4%), pp. 9367–9376.

Skills
Concepts: Machine Learning, Deep Learning, Machine Translation, Optimization, Domain Generalization.
Programming: Python, Java, JavaScript, Bash, Matlab, R, C++, Prolog.
Frameworks & Tools: PyTorch, TensorFlow, Fairseq, Faiss, Slurm, NumPy, Flask, Django, Pandas, Docker, Git, Linux, SQL,
Gremlin, Matplotlib, HTML, CSS, Jira, Confluence, LATEX.
Databases: MySQL, Oracle, JanusGraph, Cassandra, MongoDB, VoltDB, NuoDB, CockroachDB.
Languages: German (native), English (near native), Japanese (beginner).

Selected Open-Source Projects and Contributions

� � DeepOBS: Optimization Benchmarking Suite – Contributed baselines, scripts and improved software quality.
� � DomainBed: Domain Generalization Benchmarking Suite – Contributed algorithms and other features.
� � NL-Augmenter: NLP text transformations and dataset filters – Contributed filters and code quality changes.
� � Recommendation Systems: Analyzed recourse and availability under model uncertainty and discrepancy.
� App2Night: Cross-platform mobile app to create, attend, and rate user-generated events in real time.

Invited Talks and Poster Sessions

King’s College London: Virtual – Industry Insight December 2nd , 2021
Apple AI/ML Machine Translation: Virtual – Deep Learning Optimization Paper October 20th , 2021
International Conference on Machine Learning ‘21: Virtual – Spotlight Optimization Paper July 21st , 2021
KTH Royal Institute of Technology: Virtual – Deep Learning Optimization Paper September 25th , 2020
IBM Extreme Blue Conference: Cluj-Napoca, Romania – MDM Project September 3rd , 2019

Preprints and Technical Reports

2020 [4] Robin M. Schmidt and Moritz Hahn. 2020. Collaborative Filtering under Model Uncertainty. arXiv,
arxiv.org/abs/2008.10117.
2019 [5] Robin M. Schmidt. 2019. Recurrent Neural Networks (RNNs): A gentle Introduction and Overview. arXiv,
arxiv.org/abs/1912.05911.
2018 [6] Robin M. Schmidt. 2018. Improvements for the configurable Data Analysis Pipeline within a Manufacturing
Execution System. (subject to an NDA).
[7] Robin M. Schmidt. 2018. New SQL Databases: An empirical evaluation of Open Source NewSQL databases
regarding modern application scenarios. (title translated from german).
2017 [8] Robin M. Schmidt. 2017. Calculation and Evaluation of Key Performance Indicators for production within a
Manufacturing Execution System. (subject to an NDA).

Theses
2021 [9] Robin M. Schmidt. 2021. Explainability-aided Domain Generalization for Image Classification. arXiv,
arxiv.org/abs/2104.01742.
2018 [10] Robin M. Schmidt. 2018. Conception and Implementation of a Single Product Tracking System within a press
hardening production line. (subject to an NDA).

Activities and Services

Google Hash Code 2021: Organized a virtual hub and placed in the Top-15% of participating teams worldwide.
Academic Service: Served as a reviewer for ML conferences such as ICLR ’22 (highlighted reviewer) or NeurIPS ’22.
Street Photography: Samples of my side work – Selling metal, paper, or canvas prints of my street photography.