stratum: A System Infrastructure for Massive Agent-Centric ML Workloads
Summary: Stratum decouples agent planning from execution, compiling batches of Python ML pipelines into optimized graphs across heterogeneous backends, including a Rust runtime. It targets massive exploratory agentic pipeline search while preserving library compatibility, achieving up to 16.6× speedups.
(summarized by gpt-5.6-luna on Aug 28 2026)
No non-self incoming citations found for this paper in this database.
Authors
1.
Arnab Phani
(Berlin Institute for the Foundations of Learning and Data; Technical University of Berlin)
2.
Elias Strauss
(Berlin Institute for the Foundations of Learning and Data; Technical University of Berlin)
3.
Sebastian Schelter
(Berlin Institute for the Foundations of Learning and Data; Technical University of Berlin)
BibTeX Citation
@article{phani_vldb26,
title = {{stratum: A System Infrastructure for Massive Agent-Centric ML Workloads}},
author = {Phani, Arnab and Strauss, Elias and Schelter, Sebastian},
journal = {PVLDB},
series = {{VLDB} '26},
volume = {19},
number = {11},
pages = {3799--3806},
doi = {10.14778/3836663.3836726},
url = {https://doi.org/10.14778/3836663.3836726},
year = {2026}
}
Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
Rank
Citing Paper
Year
Venue
Pagerank
PreviousPage 1 / 1Next
Outgoing Citations (Sorted by Pagerank)
Showing 31 of 31 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.