Peter Albert
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPeter Albert is a versatile machine learning engineer and serial founder with eight years of experience building AI products and e-commerce ventures from San Francisco. He contributed to large-scale LLM training at MetaAI (including work on Llama 2) and improved checkpointing, CI/CD, and Azure-backed resumption workflows in the high-profile metaseq repo. Peter co-founded multiple startups—Zeta Labs/Jace AI and GetViktor—focused on applying LLMs to automate everyday web tasks and enterprise workflows, and earlier built a profitable e-commerce and board game business during university. His background combines a Molecular Biotechnology MS from TUM with hands-on backend and DevOps engineering, giving him a rare cross-disciplinary perspective on product and systems design. Pragmatic and product-focused, he moves models from research into reliable production pipelines. He is actively hiring and continues to experiment with LLM-driven automation at the intersection of web tooling and user productivity.
8 years of coding experience
7 years of employment as a software developer
Master of Science - MS, Molecular Biotechnology, Master of Science - MS, Molecular Biotechnology at Technical University of MunichBachelor of Science - BS, Bachelor of Science - BS at Leibniz Universität Hannover
Github Skills (8)
python10
cicd10
microsoft-azure9
azure9
pytorch9
logging8
testing8
data-loading7
Programming languages (2)
Github contributions (5)
facebookresearch/metaseq
Sep 2022 - Dec 2022
Repo for external large-scale work
Role in this project:
Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:56 reviews, 14 commits, 31 PRs in 2 months
Contributions summary:Peter primarily focused on refactoring and optimizing the checkpointing utilities within the `metaseq` library. They removed and renamed checkpoint flags, streamlining the saving process, and adding features for resuming training. They also implemented and tested an end-to-end test for training resumption with mocked Azure checkpoints, demonstrating their involvement in the CI/CD pipeline and cloud integration. Furthermore, the user made modifications to data loading and logging, enhancing the debugging and monitoring capabilities of the system.
transformersdockerscalelarge-scalehuggingface
Xirider/LiveCode
Jun 2020 - Aug 2021
Real-time python variable evaluation
Contributions:9 commits, 2 PRs, 6 pushes in 1 year 2 months
pythonevaluationmachine-learningvariablereal-time
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.